Why does B+ filtered power supply behave differently than a battery supply, during testing?

[CascoSieg1]

I'm an enthusiastic novice with enough tools and knowhow to be dangerous. Have built three imperfect guitar amps I am none-the-less pretty happy with, and prototyping the next one, but getting test results I just don't understand. Scroll down to "crux" of the issue if you want to skip the context...
<background>
Design requirements are basically: I'm aiming for simple to grab and go, with a few onboard fx so I don't have to keep track of pedals, built-in OD option, stereo SE EL84 power section. I also want to try some things I've haven't done yet, so, the current design is a single channel 12ax7 preamp, driving a 2-knob tone stack, with a switchable one-tube OD, into a split 12ax7 driver, into the power tubes. I'd like to implement a Sluckey trem-o-nator, and a Belton Brick for stereo reverb, but really want to build and test things out ahead of time so I can get through the build easier. So, I've breadboarded the fundamental parts of the amp and a single output channel, and all that works Ok.
I realized/discovered that the single input pre, while it's fine for cleans, wouldn't drive the OD very hard, so added a JFET (2N4416) boost to the front of that sub-circuit (a new thing for me to try), with it's power supply based on the Ceriaton 183 (http://www.ceriatone.com/ceriatone/wp-content/uploads/2019/05/OTS-183-50W-010519.jpg)... it needs tweeking but seems to work fine. That B+ filter node - the 5th node in a standard-enough power supply design - makes 25vdc (see schem attached).
Rather than using tubes for the input buffer and make up stages (one for each side) of the stereo reverb, I figured I'd use the TL072 opamps from a donor amp for those various uses, and power them with the same 25v B+ I'm making for the OD boost - 25v is overkill, but below max for those units.

<crux>
I put together one side of an opamp circuit for testing (see schem attached), and used a 25v battery (three 9-volts) since that was more convenient for my workspace at this stage, and passed a clean signal, doubling it as expected in the makeup stage. Then I fired up the amp and connected the power supply IN and GND of my circuit to the amp's 25v node and GND, instead of the battery and nothing works. So I probed for the voltage during operation at pin 8 and got 4.5v from there to GND. What the... ? So I go back and check the battery and for sure it continues to read 25v during operation. So, I check the working JFET boost circuit, and it also continues to read 25v during operation. I Also checked between pins 8 and 4 of a different TL072 - unconnected to any circuit - and got the same behavior: 25v when battery connected and 4.5 from the amp's B+.
Can anyone help me understand what's going on? Why does my voltage go down with the filtered DC supply compared to a battery?
Thanks for any help !!

 

[andrewRneumann]

Could it have to do with the over 100K of source resistance at your Node 5, versus a few Ohms for the battery?

(I admire your ingenuity BTW.)

 

[CascoSieg1]

Thanks Andy.
That sounds like more of an answer than a question, but my limited knowledge makes it unclear... (?). What you write sounds plausible, but while I don't quite understand how to work out the "why" of that, I'm thrown by the fact that the node seems to power the 2N4416 JFET just fine. So, I'm confused as to why one but not the other (?).
If it's bad idea to try to use the same B+ rail for the tubes and SS items (even though Mr. D. (RIP) did it), I could add a second smaller power transformer just for that (since I have one from the donor amp that did just that).
Thoughts?

 

[andrewRneumann]

Thanks Andy.
That sounds like more of an answer than a question, but my limited knowledge makes it unclear... (?). What you write sounds plausible, but while I don't quite understand how to work out the "why" of that, I'm thrown by the fact that the node seems to power the 2N4416 JFET just fine. So, I'm confused as to why one but not the other (?).
If it's bad idea to try to use the same B+ rail for the tubes and SS items (even though Mr. D. (RIP) did it), I could add a second smaller power transformer just for that (since I have one from the donor amp that did just that).
Thoughts?

Does the JFET draw as much current as the Op-Amp? That's not a rhetorical question... I don't work with solid state actives so I can't remember how the different components work. I would guess the op-amp is drawing more current, and that is causing your voltage drop--there is no regulation when you just use a voltage divider. When you have a source resistance of 102.7K (all those series dropping resistors in your power supply), it only take a tiny current to drop a lot of voltage. In your case, you are dropping an extra 20.5V when you hook up the op-amp. That's only 0.2mA to cause that much voltage change.

A voltage divider is great for dividing voltage. It is bad at maintaining voltage when average current changes. I don't think it is necessarily a bad idea to power SS items with B+, but they have to be devices that draw a constant amount of current (on average) if you want constant voltage.

You might be able to get it to work by reducing that 100K to a lesser value. The unloaded voltage will be higher (maybe a lot higher!), but when you connect the op-amp, it should drop to 25V if you have picked the correct value. I can't tell you the value to use, because I have no idea how much current the op-amp should be pulling if it was able to have a +25V supply. Can you simulate it?

 

[Ten Over]

A typical strategy for running op amps on a single ended supply is to configure the supply with a faux ground.


"G" is the faux ground, "F" is +15V, and actual ground is -15V. The op amp that this powers draws around 2.5mA. The maximum available current from this supply is around 3.4mA, so the voltage drop across the 68k and 51k resistors is controlled by the zener diodes and not by the op amp.


Capacitors block the actual DC voltages so that the op amp only sees the faux DC voltages.

This is another example of the same strategy:


This one is straight out of the cookbook.

 

[CascoSieg1]

@Andy,
Ah!!! I see. The data sheets show max current draws of 3ma per TL072 pair and a whopping 5ma - 15ma for the single JFET though I'd guess they draw far less in my circuit. So, it makes sense that just adding the TL072 could pull the voltage down. I am not having any luck running PSUD on my recently updated Win10 OS, so after my initial PS sim and prototype, I could never go back to see what effect the added SS components might have... and also just haven't figured out how to model transformers and rectifiers on TINA... but it's on the todo list.
@ Ten Over,
Wow! Thanks! this is just what I was looking for. I have to read up on zeners - not sure I understand them really, but I do have some around and I'm sure I can build the SS PS node you posted.

Thanks so much for the help!
~S

 

[andrewRneumann]

@Andy,
Ah!!! I see. The data sheets show max current draws of 3ma per TL072 pair and a whopping 5ma - 15ma for the single JFET though I'd guess they draw far less in my circuit. So, it makes sense that just adding the TL072 could pull the voltage down. I am not having any luck running PSUD on my recently updated Win10 OS, so after my initial PS sim and prototype, I could never go back to see what effect the added SS components might have... and also just haven't figured out how to model transformers and rectifiers on TINA... but it's on the todo list.
@ Ten Over,
Wow! Thanks! this is just what I was looking for. I have to read up on zeners - not sure I understand them really, but I do have some around and I'm sure I can build the SS PS node you posted.

Thanks so much for the help!
~S

Yeah. The Zener solution is a basically a way of setting up constant current. When the load (your SS devices) are idle, the Zeners pass more current. When the load current increases, the Zeners pass less current. So the total current being supplied remains relatively constant--and the voltage remains relatively constant. It's a fine solution to your problem, especially if dealing with relatively small currents.

Something to keep in mind though--you are designing a push-pull amp here. If these the power tubes are biased to run in Class AB, you will see some power supply sag at full drive. This sag will trickle down to the +25V supply. I'm not sure how sensitive your devices are to a sag in supply voltage, but you should plan for it. This basically means you will have to use less resistance supplying the Zeners and allow for a little more current to pass through the Zeners when at idle. I imagine you will have to build it and test it to see if you maintain +25V at full drive and then adjust from there.

Good luck! Glad you had an ah-ha moment!

 

[Old Verle Miller]

Somewhere in a galaxy far, far away this kind of thread will be looked at some many years from now and someone will be saying the beings on this planet are worth being admitted to the galactic society!

This place is a giant white board of PRACTICAL electronic engineering experience.