Some grounding questions

Lowerleftcoast

Poster Extraordinaire
Joined
Dec 30, 2019
Posts
5,623
Location
california
Does this also mean that I should tie the tail of the 5.6k grid leak at the input (V1a) to V1a's cathode, rather than to the "chassis ground point" where I have it shown now?
Correction: 5.6k should read 5.6M.

Going strictly by the book, I would say yes. For practicality, imo, no, because this should be a quiet, well filtered place in the amp.

Since the Reverberocket splits the B+ rail at the *B* node, it may be *theoretically correct* to have a second ground bus for the *D*,*E*,*F* nodes. To avoid a ground loop a resistor (most likely being bypassed with a capacitor) would be placed on one of the buses. I show the resistor circled in green below. I would appreciate @2L man to comment on this.

If I were modding the Reverberocket, I would probably add another filter node, creating some space between the first gain stage and the PI. I would place the additional filter cap near the first gain stage. I would run a single ground bus very much like you show it with the new node closest to the input/first gain stage and not bother with a second ground bus.

reverberocket11.png

(chas, I notice the input section of your layout is different than the original. Not a bad thing just different.)
 

2L man

Tele-Afflicted
Joined
Nov 23, 2020
Posts
1,530
Age
62
Location
Finland
Since the Reverberocket splits the B+ rail at the *B* node, it may be *theoretically correct* to have a second ground bus for the *D*,*E*,*F* nodes. To avoid a ground loop a resistor (most likely being bypassed with a capacitor) would be placed on one of the buses. I show the resistor circled in green below. I would appreciate @2L man to comment on this.
Sorry but I don’ t understand the purpose of resistor circled green? Unless it is ”ground loop hum kill circuit” but there should be also two diodes which current rating should be at least double the secondary fuse is and a capacitor.

That schematic is 100% Star grounding candidate ;)

I have written few times that my profession has been electronics more than 40 years and for Mains powered systems secondarys there is Zero Volts where current return to power suopply when either positive or negative operative voltage is used. When +- power supply is used only partial current flow thru 0V. Grid biased tube amps belong to +- PS group although bias current is very low.

Sometimes 0V is one bus and sometimes few wire Star.

0V acts also as a reference for signal.

Electrically Chassis has two roles, safety earth and shield but it also hold some components.
 
Last edited:

Lowerleftcoast

Poster Extraordinaire
Joined
Dec 30, 2019
Posts
5,623
Location
california
Unless it is ”ground loop hum kill circuit” but there should be also two diodes which current rating should be at least double the secondary fuse is and a capacitor.
This is why I asked you to comment. Assuming the chassis is not being used for return current. Two ground buses would need a "ground loop hum kill circuit". (The circuit would need two diodes a resistance and a capacitor.)

Let's consider a single ground bus for this project.
From your knowledge on this subject... to be theoretically correct, can the nodes be "out of order" on the ground bus? Iow, assuming the input jacks were used as the ground reference, would it matter which nodes were closer to the input jacks? The PI and the first gain stage are on node *C*. Should the node *C* attachment to the ground bus be closest to the input jacks considering node *C* services the PI?

Clearly the Reverberocket does not follow Blencowe's grounding chapter 15.4 where he states, "The amp’s input stage will always be supplied by the last filter in the
chain since this valve is the most sensitive to any noise on the power supply.
". (I appreciate Ampeg wanted more voltage for the first gain stage so they chose not to use the last filter in the chain.)


reverberocketnodes1.jpg
 

chas.wahl

Tele-Holic
Joined
Dec 8, 2012
Posts
515
Location
NYC
Correction: 5.6k should read 5.6M.
Ach! yes -- unfortunately I cannot edit to correct.
Re: whether 5.6M input grid leak should be terminated at V1a cathode resistor tail
Going strictly by the book, I would say yes. For practicality, imo, no, because this should be a quiet, well filtered place in the amp.
See the attached, (both complete PDF, and images that may not be so useful) which represent my updated schematic:
• R25 moved to V4a from V2b, to be grounded to V4 shared cathode resistor rather than Ground node E
• all grid leaks tied to the cathode resistor tails of the tube sections they serve
• additional ground bus resistor between nodes A and B, with that and the B+ dropping • resistor halved in value (1k > 2 x 500R), but wattage still not determined
• finally, an extra filter cap where I think you're suggesting it go, on B+ just before V1a (more below)
Since the Reverberocket splits the B+ rail at the *B* node, it may be *theoretically correct* to have a second ground bus for the *D*,*E*,*F* nodes. To avoid a ground loop a resistor (most likely being bypassed with a capacitor) would be placed on one of the buses. I show the resistor circled in green below. I would appreciate @2L man to comment on this.
At B+ node B, there is actually a triple fork:
- continuation without any other dropping resistor to V1b (tremolo) plate resistor
- to node D through 5k dropping resistor, and thence to E and F through additional ones
- to node C through 22k dropping resistor, and thence to V4a, V4B & V1a plate resistors
- as if that weren't enough, node F serves both V2a plate resistor, and also that of V3b, by doubling back

This is a much more complicated B+ configuration than I have seen before. I don't have a problem splitting the ground bus into additional ones, but I thought that the gathering together of grounds from the circuit sections served from each node, connecting each at the bus where the filter cap stationed at that node is connected, takes care of that -- that's what the wiring mess and back-and-forth at the bottom of the schematic (beneath the ground bus trunk that's dashed) is all about (along with some other circuiting doubling-back, which would be done under-board). Was I wrong? If so, then some guidance on how better to avoid ground loops would be appreciated. I'd like to see Merlin Blencowe diagram his current flow on this schematic!

I'm not sure I understand why the resistor would help, and what its value and wattage, and the value and voltage rating of an associated bypass cap would be.
If I were modding the Reverberocket, I would probably add another filter node, creating some space between the first gain stage and the PI. I would place the additional filter cap near the first gain stage. I would run a single ground bus very much like you show it with the new node closest to the input/first gain stage and not bother with a second ground bus.
I think I get it -- poor little sensitive V1a is hung off the node C, with only 3 filter caps between it and the dreaded AC ripple, right? But what would be a good value for an additional filter cap situated in the circuit between node C and V1a?
(chas, I notice the input section of your layout is different than the original. Not a bad thing just different.)
I don't think it's actually different, circuiting wise -- I believe I have all the same inputs (guitar, mic, accordion) with the same values they have in the original schematic, only I've swapped the accordion and mic (inputs 2 & 3) to be 3 and 2 instead, just to make it easier to draw and implement hanging all those resistors off the V1a grid pin rather than putting them on the board -- I'd use a terminal strip to help manage that. I'm pretty sure that's what EL34world's Sonny Reverb did on his Verberwaffe, putting his V1 tube on the input jack side of the board (with ground bus and controls) rather than on the output jack side, like all the other tubes -- though he only uses one input jack:
1656971899878.png

I'm more familiar with the input grid leak being connected on the jack side of the grid stoppers between jacks and V1a, but I guess it doesn't make a difference, since Ampeg shows it that way -- I've just connected all 3 grid stoppers and the grid leak to the tube section's grid pin.

1656977420118.png


1656977643368.png


1656977769194.png
 

Attachments

  • 1656973784392.png
    1656973784392.png
    514.2 KB · Views: 16
  • 1656973904985.png
    1656973904985.png
    227.7 KB · Views: 12
  • 1656973993587.png
    1656973993587.png
    325.6 KB · Views: 13
  • Reverberocket-R12R-schem-2022-07-04.pdf
    443.8 KB · Views: 6
Last edited:

Lowerleftcoast

Poster Extraordinaire
Joined
Dec 30, 2019
Posts
5,623
Location
california
poor little sensitive V1a is hung off the node C, with only 3 filter caps between it and the dreaded AC ripple, right? But what would be a good value for an additional filter cap situated in the circuit between node C and V1a?
Probably reduce the 22k at the C node to 18k and add a 3.9k before the new filter cap (20uF). The new (3.9k/20uF) RC filter corner frequency is ~2Hz. The RC filter corner frequency at C (18k/20uF) will hardly change.
additional ground bus resistor between nodes A and B, with that and the B+ dropping • resistor halved in value (1k > 2 x 500R), but wattage still not determined
Blencowe shows an additional cap. The resistance is similar to a sag resistor. Essencially that is what you will have. The 1k stays as the first dropping resistance.
If you want to do this, the total resistance would be maybe 100R. (With the little PT... 50R to 75R???) Wattage rating... 2 x 25R 5W. 2 x 35R 6W.
Was I wrong?
I think you are starting to get this. Ideally the wire runs for the grounds should be kept short. (You want to avoid the Fender style long run to cap pans, for instance.)
I'd like to see Merlin Blencowe diagram his current flow on this schematic!
Me too. hah.
I'm not sure I understand why the resistor would help, and what its value and wattage, and the value and voltage rating of an associated bypass cap would be.
When there is a ground loop, the electrons have roughly the same resistance for either path. Adding a ”ground loop hum kill circuit” will discourage one of the paths. The resistor is small 10R to 25R.
 

chas.wahl

Tele-Holic
Joined
Dec 8, 2012
Posts
515
Location
NYC
Yikes! I'd be lying if I pretended that I'm not getting out of my comfort zone on suggested improvements to the Ampeg circuiting.
Probably reduce the 22k at the C node to 18k and add a 3.9k before the new filter cap (20uF). The new (3.9k/20uF) RC filter corner frequency is ~2Hz. The RC filter corner frequency at C (18k/20uF) will hardly change.
I'll consider this, but I'll have to confer with my (in @King Fan parlance) amp-sensei (not that you guys aren't, but he's related to me).
Blencowe shows an additional cap. The resistance is similar to a sag resistor. Essencially that is what you will have. The 1k stays as the first dropping resistance.
If you want to do this, the total resistance would be maybe 100R. (With the little PT... 50R to 75R???) Wattage rating... 2 x 25R 5W. 2 x 35R 6W.
This I'm pretty much flummoxed by, I have to admit. I understand (now, I guess) that Blencowe's diagram shows something like a filter made up of two caps in parallel -- think 5F2-A circuit -- but with small-value dropping resistors between them on both B+ and Ground sides. But I'll have to think that over -- what with the idea of adding another filter between B+ node C and V1a, I'd be adding two more large caps to the circuit (for a total of 8!) and that's a big leap.
I think you are starting to get this. Ideally the wire runs for the grounds should be kept short. (You want to avoid the Fender style long run to cap pans, for instance.)
Thanks for saying so, but I feel like I'm getting into deep water. I need to try to wrap my head around where the ground loops actually are in the circuit I posted just previously, and how those might be avoided. Also, I've read but don't really follow Blencowe's advice here:

1657025244452.png


What ground impedance is there (the resistive element shown dashed in the diagram), really, in a length of ground bus between nodes?

When there is a ground loop, the electrons have roughly the same resistance for either path. Adding a ”ground loop hum kill circuit” will discourage one of the paths. The resistor is small 10R to 25R.
I get that, but I am having trouble, as I said, visualizing where the loops are. Need to think more, but have other things to do today, and will have to mull subconsciously only.

Thanks, though. As I said in the beginning of my other thread about the Reverberocket, I think it's a very interesting circuit, that's different from the typical Fenders being discussed mostly on TDPRI, and just cogitating about those differences is a sort of education.
 

King Fan

Poster Extraordinaire
Ad Free Member
Joined
Jan 1, 2013
Posts
8,404
Location
Salt Lake City
I admire your journey here, @chas.wahl . Me, I’m a simple clone-minded guy. If the Ampeg works, in fact iconically, I’d be tempted to just translate their schematic / layout onto a more Fender-y (and roomy?) board. Sure, I’d modernize and upgrade where safety or longevity could be improved, but I might not Merlinize the whole ground scheme. Just me. And as you say, this 'what if' kind of thinking is a great way to learn.
 
Last edited:

2L man

Tele-Afflicted
Joined
Nov 23, 2020
Posts
1,530
Age
62
Location
Finland
This is why I asked you to comment. Assuming the chassis is not being used for return current. Two ground buses would need a "ground loop hum kill circuit". (The circuit would need two diodes a resistance and a capacitor.)

Let's consider a single ground bus for this project.
From your knowledge on this subject... to be theoretically correct, can the nodes be "out of order" on the ground bus? Iow, assuming the input jacks were used as the ground reference, would it matter which nodes were closer to the input jacks? The PI and the first gain stage are on node *C*. Should the node *C* attachment to the ground bus be closest to the input jacks considering node *C* services the PI?

Clearly the Reverberocket does not follow Blencowe's grounding chapter 15.4 where he states, "The amp’s input stage will always be supplied by the last filter in the
chain since this valve is the most sensitive to any noise on the power supply.
". (I appreciate Ampeg wanted more voltage for the first gain stage so they chose not to use the last filter in the chain.)


View attachment 1000973
I think the order how the B03, B04, B05... "grounds" are connected to single common gound bus does not have any adverse effect in audio amplifiers. Perhaps RF circuits there might be more importance but if I have learn something of it I have forgot because its been 33 years I quit working micro wave systems.

Chas, I think on that kind filtering the power transformer CT is not yet Common ground because there is a resistor and because of resistance the CT voltage go negative against ground. Common ground start, or better saying finishes, to B+1 filter capacitor negative B01.
 
Last edited:

Lowerleftcoast

Poster Extraordinaire
Joined
Dec 30, 2019
Posts
5,623
Location
california
Yikes! I'd be lying if I pretended that I'm not getting out of my comfort zone on suggested improvements to the Ampeg circuiting.
I hear ya. Usually an amp can be quiet in stock form. @King Fan and I have agreed many ground schemes can be quiet no matter which *rules* have been broken. It seems the key is to keep noise away form the early gain stages, and to have sufficient filtering from the reservoir and first filter caps.

Imo, the lack of filtering on the first gain stage was a cost saving measure. I would change that unless this is to be a strict clone. Depending on how you play, I might add more uF to the reservoir and first filter cap to tighten bass. (It might reduce hum as well.)

Like 2L man, I endorse not using the chassis for current flow, just using one connection as a ground reference.
what with the idea of adding another filter between B+ node C and V1a, I'd be adding two more large caps to the circuit (for a total of 8!) and that's a big leap.
So many amps do not use Blencowe's resistor in the return path, imo you can go without that mod. (Iirc 2L man has written of a project using several equivalent resistances on the return path.)
If space is a concern on the *C* node, a 10uF with a 3.9k (3.9k/10uF) RC filter would have a corner frequency is ~4Hz. That would reduce 120Hz ripple by over 30dB at the first gain stage. Only 6dB less reduction than the 20uF cap would provide.
I need to try to wrap my head around where the ground loops actually are in the circuit I posted just previously, and how those might be avoided.
The ground loop I was referring to was having two ground buses. I am not too concerned with how your layout handles the return current.
What ground impedance is there (the resistive element shown dashed in the diagram), really, in a length of ground bus between nodes?
If I am correct, the impedance would come from interaction of the ground bus with outside influences creating capacitance and/or inductance. This is why wire runs should be short.
I get that, but I am having trouble, as I said, visualizing where the loops are.
The obvious ground loop was on the marked up schematic I provided, suggesting two ground buses. I show the loop in green below. On the right, out of view, is where the two ground buses connect at the *B* node.

958232-7bf9d30.png

As I said in the beginning of my other thread about the Reverberocket, I think it's a very interesting circuit, that's different from the typical Fenders being discussed mostly on TDPRI, and just cogitating about those differences is a sort of education.
I agree. I am glad you brought it up.
 

2L man

Tele-Afflicted
Joined
Nov 23, 2020
Posts
1,530
Age
62
Location
Finland
I second Lowerleftcoast observation and think that the 500 ohm resistor on Common return between A and B might not function because B is aleady Screen return! Pulsating current will make reference between OT/plate and screen feed irrational, and stages earlier too, when there comes voltage loss to this resistor.

Merlins idea is to improve power supply filtering installing resistors to current return before the B+1 filter but also increase filter capacitor count. I have used same method in many amps but always used solid wire(s) between B01, B02, B03...

I have used many, usually 3, 4 or even 5 RCR filters where capacitors are smaller. However I can't say is it because of resistors on return current what make amp hum less? Possibly multiple RC filters would have same effect? But if capacitors are wound to turretts it is as easy to install resistors to return as it is to install them to feed.
 
Last edited:

2L man

Tele-Afflicted
Joined
Nov 23, 2020
Posts
1,530
Age
62
Location
Finland
Is there a need for "humdinger" when tubes cathodes are indirect heated? Usually it is used when tube AC heated filament act as a cathode (direct heating).

Perhaps better effect comes if you elevate filaments higher than about 20VDC what the power tube cathodes come? If you install a capacitor bleed resistor installing one more bleed series resistior and there comes voltage divider where filament CT (or artificial CT) could be connect.
 
Last edited:

chas.wahl

Tele-Holic
Joined
Dec 8, 2012
Posts
515
Location
NYC
I've modified my schematic, provisionally, to
• include the extra filter cap C-31 between B+ node C and the input triode V1a as suggested by @Lowerleftcoast, called node C-1, to which V1a is grounded except for the Volume pot lug 1 grounded to node F (V2a, the next stage); and to
• add an additional filter cap before B+ node A (that is, new reservoir cap, also suggested by @Lowerleftcoast) and connected to node A filter cap (increased in value to 30 or more uF) through a choke.

The choke is shown as a pretty big one, Hammond 194D: 5 H, 150 mA, 107 Ω DCR -- I don't know if the smaller 194B (90 mA) would do the job; need to investigate that.

1657149320888.png
 

Attachments

  • Reverberocket-R12R-schem-2022-07-05.pdf
    440.9 KB · Views: 7

Lowerleftcoast

Poster Extraordinaire
Joined
Dec 30, 2019
Posts
5,623
Location
california
include the extra filter cap C-31 between B+ node C and the input triode V1a as suggested by @Lowerleftcoast, called node C-1
This needs to be an RC filter, so add a 3.9k dropping Resistor anywhere in the wire between the 22k dropping resistor and the new C-1 cap. Since this new 3.9k dropping resistor is being added, the 22k should change to 18k so the voltage at V1a does not change.
The choke is shown as a pretty big one, Hammond 194D: 5 H, 150 mA, 107 Ω DCR -- I don't know if the smaller 194B (90 mA) would do the job
Since the choke is before the OT, all of the current for the amp goes through the choke. 90mA is too little, imo. If the choke were placed after the OT, like most Fender designs, a 90mA choke would be sufficient.
 

chas.wahl

Tele-Holic
Joined
Dec 8, 2012
Posts
515
Location
NYC
This needs to be an RC filter, so add a 3.9k dropping Resistor anywhere in the wire between the 22k dropping resistor and the new C-1 cap. Since this new 3.9k dropping resistor is being added, the 22k should change to 18k so the voltage at V1a does not change.
Is there a rationale behind choosing 4k for the C-1 dropping resistor? I didn't go to RC or LC school yet, I'm ashamed to admit. Does it make any sense to adjust the value of the 47k plate resistors of the PI triodes, so that the dropping resistor plus the plate resistor still amounts to 69k?
Since the choke is before the OT, all of the current for the amp goes through the choke. 90mA is too little, imo. If the choke were placed after the OT, like most Fender designs, a 90mA choke would be sufficient.
May I presume that there's a greater benefit, in terms of ripple reduction, to feeding the OT from node A rather than directly from the Reservoir cap? If there is, I don't mind paying the extra $20 for the upgrade from 194B to 194D.
 

andrewRneumann

Tele-Afflicted
Joined
Mar 22, 2020
Posts
1,981
Location
Cincinnati, OH, USA
Is there a rationale behind choosing 4k for the C-1 dropping resistor? I didn't go to RC or LC school yet, I'm ashamed to admit. Does it make any sense to adjust the value of the 47k plate resistors of the PI triodes, so that the dropping resistor plus the plate resistor still amounts to 69k?

May I presume that there's a greater benefit, in terms of ripple reduction, to feeding the OT from node A rather than directly from the Reservoir cap? If there is, I don't mind paying the extra $20 for the upgrade from 194B to 194D.

Why did you end up with a choke? I must be behind in this thread. A choke before the OT is usually not necessary in a push-pull because the ripple hum on each power tube gets cancelled.
 

Lowerleftcoast

Poster Extraordinaire
Joined
Dec 30, 2019
Posts
5,623
Location
california
Is there a rationale behind choosing 4k for the C-1 dropping resistor? I didn't go to RC or LC school yet, I'm ashamed to admit.
Nothing to be ashamed about. I am still learning. I have no formal training. My informal training came from my father. I wasn't very interested in learning about electronics but I grew up with his hobbies/work all around me. At a young age he would have me tear down gear and install new parts. Most school projects involved building something with buzzers and lights. I even built a p2p 11 transistor radio from a schematic and scrounged parts for a 7th grade science project. (In those days the transistors were germanium.) Most of what I have learned was from osmosis. I now lament I should have paid better attention. The one thing I remember him saying the most is "close enough, it will work" when I didn't have the exact value of cap or resistor. My young self did not like seeing a schematic with written values and my father saying those values can be altered.

The way I look at these amplifier circuits, they are all just variations of the same or very similar thing. RCA put out several schematics to sell tubes and folks put together copies and "close enough, it will work" variations to make radios and amplifiers.

Over my shoulder, I still hear "close enough, it will work".

When I suggested adding the low pass RC filter, my thought process...
C1 could be 20uF like all of the other filter caps in this amp.
I don't want to change the character of the amp but there should be adequate filtering at this new node.
I have to keep the same voltage at V1a or the character of the amp may change. To accomplish that, the new dropping resistor along with the 22k dropping resistor should equal about 22k. I know I can get away with lowering the value of the 22k dropping resistor a little without effecting the PI.
How much resistance must I use to make an adequate low pass RC filter? To keep from doing the math, I use an online RC filter calculator. A First Order RC filter rolls off 6dB per octave above the cutoff frequency. The loudest frequency we want to get rid of is 120Hz. (The frequency of the rectified AC. Ripple current.) If I want to take off 36dB, the cutoff frequency would be 6 octaves below 120Hz. (So half of 120 = 60... half of 60 = 30... etc for 6 octaves.) I'd like to have a cutoff frequency of about 2Hz. Plug in 2Hz and 20uF into the RC filter calculator... = 3981R ~ 3.9k. If the new dropping resistor is 3.9k, the 22k dropping resistor should reduce to ~18k. I check the RC filter calculator to compare the cutoff for the original 22k/20uF and 18k/20uF, there is hardly any difference. I estimate the approximate voltage drop across the 18k for the PI. "Close enough, it will work".
Why did you end up with a choke?
I am going to agree with andrew. Imo, a choke will change more in the reverberocket than just the hum level. Leave space for it if you wish, but try it first without.
 

chas.wahl

Tele-Holic
Joined
Dec 8, 2012
Posts
515
Location
NYC
I'm out of my depth here, but my instinct is that trying to "improve" the first node filtering, and maybe even adding more filtering in the odd circuiting that forks off the B+ at the PI and sends it directly to the input triode, are too much for me to deal with.

Regarding the first, I take a look at something like the 6G3 (Brown Deluxe) and see that it's got a single 16 uF cap as a reservoir. The venerable AB763 Blackface Deluxe Reverb has no reservoir cap at the node serving the OT. The 5F2-A goes to 32 uF (flirting with the boundary reservoir loading for a 5Y3GT), but it's a single-ended amp with no ripple canceling inherent in the push-pull power stage. These are near-contemporaries, and people build clones of them pretty much faithfully time and time again.

Regarding the second, filtering of the B+ occurring before the input stage, and again looking at a 6G3, I see that it has filtering of 56 uF total. The AB763 has only 48 uF total. Ampeg's schematic has 60 uF before the input stage. That seemed to work for Everett Hull and Jess Oliver, at the time (1961-2) . . . but . . .

The next iteration of the Reverberocket (R-12-RB, maybe late 1963-ish) has changed the rectifier (5Y3 gone, SS rectifier in), the OT has a different number, changed the output tubes to 7591s, and they upped the reservoir cap to 60 uF, with another 40 at the PI -- so 100 total. They never used a choke, that I'm aware of; point taken. Whether this overhaul was due to ripple/hum problems or the desire to "clean it up" with the 7591s is unclear. Something about the amp, after its introduction, prompted all this.

So: I'm going to proceed with layout, trying to leave space for a couple additional filter caps. If no choke is recommended, I'm back to trying to figure out the value of paired dropping resistors that will protect the rectifier, but involve minimum voltage drop to the whole B+ chain beyond that additional cap. One simpler change might be to increase the reservoir cap C1 value to 30 or 33 uF, and trust that the 5Y3 can deal with that.

@Lowerleftcoast: you didn't respond to my question about whether, if I were to insert another filter before V1a, to increase the value of the PI plate resistors from 47k to 51k, so that the PI plates see the same 69k as they did before the 22k dropping resistor was lowered to 18k.
 

2L man

Tele-Afflicted
Joined
Nov 23, 2020
Posts
1,530
Age
62
Location
Finland
My knowledge of Chokes is very limited but Chokes have voltage, inductance and current ratings and together with filter capacitance the inductance improves the filtering. When Choke is placed between rectifier and first filter capacitor there comes more voltage drop but filtered voltage stay better requlated against current changes but hadly any instrument amp use Choke there.

Most common place is between OT feed and Screen feed where passing current is significantly lower. After first filter before OT feed is where Choke is used most in SE amps.

Chas, now that your PT is DC rated for 90mA and Choke is 90mA I think you could use or at least test Choke before the OT feed? There it would filter the voltage for a current which is needed for quite high clean power. For full power the Choke might saturate during very high current peaks but then sound would be distorted by few other mechanisms as well :)

Perhaps a DPDT switch could be installed to switch Choke between pre and post OT feed?
 

andrewRneumann

Tele-Afflicted
Joined
Mar 22, 2020
Posts
1,981
Location
Cincinnati, OH, USA
@chas.wahl if I understand correctly you are reconfiguring the HT feeds to the preamp and you’re worried you may not have enough filtering and voltage at V1A. Do you know what the original B+ or plate voltage on V1A was? Out of all the preamp valves, V1A is can be at quite a low voltage because it doesn’t swing much as the input voltage from the guitar is so tiny. Because the swing is so low, V1A is usually the most linear stage. So V1A’s purpose is just to make the signal louder and doesn’t contribute much to tone. (I can hear many gasps from the audience right now.) My point is that your can probably stand to drop some voltage at V1A with a standard RC filter with minimal impact on the performance.
 

Lowerleftcoast

Poster Extraordinaire
Joined
Dec 30, 2019
Posts
5,623
Location
california
The venerable AB763 Blackface Deluxe Reverb has no reservoir cap at the node serving the OT.
Yes it does... highlighted below 32uF (2 x 16uF).

InkedDRAB763.jpg
Regarding the second, filtering of the B+ occurring before the input stage, and again looking at a 6G3, I see that it has filtering of 56 uF total. The AB763 has only 48 uF total. Ampeg's schematic has 60 uF before the input stage.
The total uF is not important. The key is the low pass filter(s). They are important. Each RC filter drops 6dB of 120Hz ripple per octave over the cutoff frequency. The resistance and the capacitance determine the cutoff frequency. If the filters only reduce the 120Hz ripple by 24dB everyone will hear the hum.
Reducing the 120Hz in stages is the standard design. The 1st stage will reduce the 120Hz. The 2nd stage will be reducing the much smaller 120Hz content and the next stage(s) will reduce the even smaller 120Hz content to where it cannot be heard. (It is kinda like compound interest, each day (stage) increases the earnings adding in what was earned the day before.)
The next iteration of the Reverberocket (R-12-RB, maybe late 1963-ish) has changed the rectifier (5Y3 gone, SS rectifier in), the OT has a different number, changed the output tubes to 7591s, and they upped the reservoir cap to 60 uF, with another 40 at the PI -- so 100 total. They never used a choke, that I'm aware of; point taken. Whether this overhaul was due to ripple/hum problems or the desire to "clean it up" with the 7591s is unclear. Something about the amp, after its introduction, prompted all this.
Chances are Ampeg was moving with the times, trying to carve a niche, and trying to reduce cost.
Tube rectifiers, and expensive, unreliable large Selenium diodes were being replaced with new Silicon diodes. Tube radios were being replaced with SS transistor radios. There was a lot of hype going on at that time for the new *space age* technology. Capacitance was cheaper as well and the SS rectifier was not bound to 4uF to 60uF constraints of tube rectifiers. Cheaper more filtering and possibly a cheaper PT... win-win-win.

Chokes were/are expensive compared to other filters. The big advantage... the LC filter is a 2nd order filter. 2nd order filters drop 12dB per octave. 12 dB per octave can be very desirable in the early stages, remember the compounding of later stages.

As 2L man noted the choke changes other things as well. I am not convinced a choke in a Reverberocket would be wanted. OMMV. (If you take 2L man's advice, putting the choke on a switch, be very careful. There will be high voltage on that switch.)
you didn't respond to my question about whether, if I were to insert another filter before V1a, to increase the value of the PI plate resistors from 47k to 51k, so that the PI plates see the same 69k as they did before the 22k dropping resistor was lowered to 18k.
As my father would say... "close enough, it will work".

Don't worry about it. The tolerances of the components are looser than that.

___________________________________________________________

It is your build and your choice but I hope you install the *C1* RC filter. It will solve the odd ground scheme wiring and has the bonus of providing more filtering. I see no downside to this mod. I view it as correcting something that should have been done in the first place.
 
Last edited:




Top