Sizing a Power Transformer From Scratch

Discussion in 'Shock Brother's DIY Amps' started by andrewRneumann, Sep 29, 2020.

  1. andrewRneumann

    andrewRneumann Tele-Holic

    Posts:
    591
    Joined:
    Mar 22, 2020
    Location:
    Cincinnati, OH, USA
    Hi All,

    Slowly making headway on designing the 2xEL84 Class AB1 amp with twin cathode follower preamp. My only experience with picking power transformers is Class A single ended designs, so I’ve got questions.

    Here’s what I have so far as design estimates:
    B+1 voltage 400V @ idle
    B+1 current 55mA @ idle, 201mA peak @ power tube full drive

    1st question, does the idle current vs. peak current at full drive look about right for P-P EL84’s howling at 400V plates and 320V screens? For the peak overdrive condition, I just added up the estimated plate and screen current for one tube (since the other tube is cutoff) and then added in estimates for current from the preamp tubes.

    01BED79B-D18E-4149-93D4-7FA8C8AADDE5.jpeg

    Transformer: Hammond 373EX
    Rectifier: 5AR4/GZ34

    I went through Merlin’s power supply design using the above conditions (idle and full overdrive). I estimated B+1 at 408VDC at idle, sagging to 384VDC at full overdrive. I estimated the maximum overdrive RMS ripple current in the input filter cap (reservoir) to be 362mA.

    Since this is a center-tapped HT winding and I’m using a two-phase rectifier, Merlin notes that actual RMS ripple current in one-half of the secondary is 0.707 times the full amount, so that brings the RMS ripple current seen by each half of the secondary winding down to 256mA.

    The Hammond 373EX secondary is rated at 325v-0-325v @345mA. So it appears this transformer can handle my demands.

    My second question is, is the power transformer too large for my needs? Is the current rating listed on the Hammond datasheet a hard limit that I should not exceed? Or can I exceed it and just expect more sag out of the transformer. Maybe a smaller PT would sag more and add a nice dynamic to the overall feel of the amp. If so, how small could I safely go on the current rating, assuming the HT winding still provided me with 400V B+1 at idle?

    I was surprised that I would need such a big power transformer for this amp with 2xEL84’s. But I’m new to this particular game, so maybe that’s just the way it is.

    Any tips or points about what I might be missing? I searched the web on this and was surprised to find a lot of people just met the current demands at idle and made no estimates of what the current would rise to when driving the amp hard. But I’m assuming in the P-P world, you have to look at worse case current and that would be when overdriving the power tubes. Correct?
    794DB184-5A6F-404D-AA9B-3E4A706F4B35.jpeg
     
    NSB_Chris likes this.
  2. Tom Kamphuys

    Tom Kamphuys Tele-Holic

    Age:
    41
    Posts:
    606
    Joined:
    Sep 18, 2018
    Location:
    The Netherlands
    andrewRneumann likes this.
  3. elpico

    elpico Tele-Afflicted

    Posts:
    1,124
    Joined:
    Sep 14, 2011
    Location:
    Vancouver BC
    The 373EX is larger than Hammond's AC30 power transformer. I mean, it'll work, but it is overkill. (and 5 pounds heavier than you really need)

    150ma should be plenty.
     
    Nickfl and andrewRneumann like this.
  4. tubegeek

    tubegeek Friend of Leo's

    Age:
    60
    Posts:
    3,770
    Joined:
    Jan 31, 2020
    Location:
    Brooklyn, NY
    I *think* one issue is that you're not generally running at max overdrive for extended periods of time, so the transformer doesn't suddenly overheat, that takes a bit of time to develop.

    Instead what you *probably* encounter more prominently is the voltage sag issue, and I guess the smaller the transformer rating, the more sag you'll get. At some point you will reach a point where the transformer is so stressed by the overdrive demands that it will become unreliable and fail.

    Then you install the next size up. It's like a $50 fuse.

    ;)
     
    Last edited: Sep 30, 2020
    andrewRneumann likes this.
  5. Tom Kamphuys

    Tom Kamphuys Tele-Holic

    Age:
    41
    Posts:
    606
    Joined:
    Sep 18, 2018
    Location:
    The Netherlands
    I don't like doing things just because they've always done it like that either. I would like to understand it.

    My first post wasn't any help in that area, other than an indication that your line of thought probably is not correct.

    Here is a different approach:
    In your other thread you state 400V, 25W for the two el84's. That would result in only 63mA. Add a few mA for the other tubes and that's it.
     
    andrewRneumann and tubegeek like this.
  6. Phrygian77

    Phrygian77 Friend of Leo's

    Posts:
    2,228
    Joined:
    Apr 30, 2016
    Location:
    Crawfordville, FL
    That transformer is probably just big enough to drive a quad of 6L6s.

    Look at Hammond's 'Design Guide For Rectifier Use'. They hint in that document that the AC current rating already accounts for some ripple current...

    "Use the formulas below as a guide, shown for common D.C. supplies. Included in the formulas higher peak to peak capacitor charging current in the filter. "

    https://www.hammfg.com/electronics/transformers/rectifier?referer=968
     
    andrewRneumann likes this.
  7. Tom Kamphuys

    Tom Kamphuys Tele-Holic

    Age:
    41
    Posts:
    606
    Joined:
    Sep 18, 2018
    Location:
    The Netherlands
  8. Lowerleftcoast

    Lowerleftcoast Friend of Leo's

    Posts:
    2,059
    Joined:
    Dec 30, 2019
    Location:
    california
    @Tom Kamphuys
    My nerd education is not enough to be understanding this. Andrew said he was designing a class AB1.
    I will ask this: In class AB1 the screen grid should have no current, correct? So there should be no 9mA (screen)?
     
    andrewRneumann likes this.
  9. andrewRneumann

    andrewRneumann Tele-Holic

    Posts:
    591
    Joined:
    Mar 22, 2020
    Location:
    Cincinnati, OH, USA
    Thanks everybody for chiming in. I had a feeling that xfmr was way too big, but needed some backup.

    I think that sounds about right. I went back and used an estimate of RMS current at full drive instead of peak. Essentially, just multiplying the peak by 0.707, even though I know this is just a thumb suck. It came in at around 175mA. The next size down

    Yeah, that's what I am thinking--so what do we design for? It seems like a gray area--unlike SE where you have a relatively constant current no matter the drive level. As noted above, I went ahead and tried using an RMS current (a wild assumption) at max drive. That reduced the RMS ripple current in the transformer down to about 175mA, so that can be a much smaller xfmr. (The 376X... versus the 373EX it is less than half the weight.)

    Ha! More like $125! I don't want that to happen--I will be installing fuses on the HT for sure.

    I'm with you Tom, that's why I get into the math and do a lot of things by hand (back checked with the computer if I can). Do it enough times and it actually starts making sense. Still... this particular stuff is not always making sense... yet.

    Your approach is simple, but maybe too simple. Merlin talks about "apparent power"--the total power the transformer is handling--and "real power"--the actual work done in the load. There's a pretty big difference between the two. If I did my calculations right, at full drive, the transformer has 70% more current than the current actually delivered to the load. So maybe 70% more than 63mA?

    At idle... but I need to bump it up for driving the power tubes, right? I am going to idle them a little cooler than that, but it seems like with a P-P amp you don't really care about the idle current for sizing the PT--it's the least amount of current the xfrmr will have to supply.

    Yeah, it was too big. Can you explain how to use those formulas? I looked at those and got totally confused. Either they give me a voltage that seems way to high or way too low, but I'm probably misusing them.

    Here's the applicable drawing. I'm looking at the 376X which has Vsec of 640v under load or 694.5v with no load. The current rating on the secondary is 173mA.

    full-wave-capacitor-input-load.jpg
     
    Last edited: Sep 30, 2020
  10. andrewRneumann

    andrewRneumann Tele-Holic

    Posts:
    591
    Joined:
    Mar 22, 2020
    Location:
    Cincinnati, OH, USA
    It means operation with little or no control grid current, if I understand it correctly. I.e. the tube clips at Vg1=0V. Someone please correct me if I'm wrong though!
     
    tubegeek and Lowerleftcoast like this.
  11. andrewRneumann

    andrewRneumann Tele-Holic

    Posts:
    591
    Joined:
    Mar 22, 2020
    Location:
    Cincinnati, OH, USA
    Ok round 3. I didn’t like the 376X because I was getting B+ somewhere between 420V and 428V. 400V was my absolute max for the EL84’s and I promised myself if anything I would come in below that, but not over it. I could have knocked down the voltage with the 376X by putting in some diodes and series resistors between the transformer and rectifier, but why waste the power? Plus, increasing the source resistance seen by the rectifier would increase the amount of sag quite a bit. Plus, plus, if I can keep the peak unloaded voltage below 450V, I can safely use 450V caps in the power supply filtering network. So I reran the calculations for a the 372DX, which has a 648v (unloaded) center-tapped secondary rated for 144mA.

    Summary

    Transformer Hammond 372DX
    Rs = 153R (seen by one plate of 2-phase rectifier)

    Rectifier GZ34
    Rs-idle = 127R
    Rs-max = 87R

    Total Source Resistance
    Rs-idle = 280R
    Rs-max = 240R

    Load

    Iidle = 55mA (DC)
    Imax = 146mA (RMS)

    Equivalent Load Resistance
    Rl-idle = 8050R
    Rl-max = 2898R

    Reservoir Capacitance
    47 uF

    Factors
    fCRl-idle = 22.7
    fCRl-max = 8.2

    Rs/Rl-idle = 0.035
    Rs/Rl-max = 0.083

    From Merlin’s Charts:
    DC Output % of Unloaded RMS AC Voltage
    Idle = 119% => B+ at idle = 1.19 x 324 = 386V
    Max = 109% => B+ at max = 1.09 x 324 = 353V

    Ripple Current Ratio at Max Signal
    RMS = 1.65 => RMS current = 1.65 x 146mA = 241mA
    Peak = 3.4 => Peak current = 3.4 x 146mA = 496mA

    The RMS current in each leg of the secondary winding and each plate of the rectifier is 0.707 times the total current.

    RMS current in each half of secondary winding = 241mA x 0.707 = 170mA, but Hammond doesn’t give a limit on this as far as I know. The only limit is a generalized DC output—144mA for the 372DX.

    I based my estimated currents on a constant B+ of 400V, and this analysis reveals that the voltage is 386V sagging to 353V. So my load current estimates are probably a bit on the high side, especially the “max” current since it didn’t account for sag. I will probably have another look at the power tube design and see if I might decrease the screen resistor size since I have a double whammy of power supply sag and 1.5k on the screen.

    I’m going to interpret the Hammond secondary current rating as an average DC current supplied by the power supply, so if I’m at 146mA RMS at full drive (most likely lower due to sag) then the 372DX 144mA rating is right where I need to be. Hopefully I have interpreted that right.

    As far as these equations in Hammond’s literature, if I use peak-peak voltage for secondary vac (Sec. V A.C.), the V (Avg) D.C. equation actually make some sense. Why Hammond would switch to peak-peak for this equation boggles my mind.


    A3F22E5F-5F36-40DE-8F65-5A91ED4663F2.jpeg

    For the 300-0-300 secondary, the peak to peak is 600*1.414 = 848.4v

    So V (Avg) D.C. = 0.45 x 848.4 = 382V ... pretty close to the 386V that I calculated! But again, I have no idea if I’m doing that the way Hammond intended. Using peak to peak voltage makes no sense to me. If had to guess I would say that 0.45 was a typo and should be 0.64.

    So... Hammond 372DX sound like a reasonable choice?
     
  12. elpico

    elpico Tele-Afflicted

    Posts:
    1,124
    Joined:
    Sep 14, 2011
    Location:
    Vancouver BC
    Last edited: Oct 1, 2020
  13. POS Guitars

    POS Guitars Tele-Meister

    Posts:
    185
    Joined:
    Sep 9, 2019
    Location:
    Los Angeles
    Nice transformer. We make our own by hand with vintage combs. The Hammond stuff is just too rich for me:)
     
  14. Tom Kamphuys

    Tom Kamphuys Tele-Holic

    Age:
    41
    Posts:
    606
    Joined:
    Sep 18, 2018
    Location:
    The Netherlands
    No it is not at idle. It is the average current over the sine.

    Have you seen this: http://www.valvewizard.co.uk/smoothing.html ?

    It's not the most obvious location, but there is a nice example determining the total current of a class AB push-pull amp.
     
    andrewRneumann likes this.
  15. andrewRneumann

    andrewRneumann Tele-Holic

    Posts:
    591
    Joined:
    Mar 22, 2020
    Location:
    Cincinnati, OH, USA
    Oh my bad. It must have been for just one tube? So double that and add in the other tubes? If so, that comes out pretty close to my 146mA hack.
     
  16. schmee

    schmee Doctor of Teleocity Silver Supporter

    Posts:
    12,275
    Joined:
    Jun 2, 2003
    Location:
    northwest
    Other than all the tech stuff, a PT can make a small amp clean or gritty. At higher volumes. I used to think as long as it supplied enough amps why would it matter? Then I put a big PT in a Princeton Reverb and the headroom went out of sight, as well as it was a dead quiet amp.. At that point I became a believer.
     
  17. andrewRneumann

    andrewRneumann Tele-Holic

    Posts:
    591
    Joined:
    Mar 22, 2020
    Location:
    Cincinnati, OH, USA
    This thread has helped me distill some concepts so thanks everybody. Ultimately, I guess my questions amounted to figuring the full drive current correctly and then picking the right sized PT to go with. The voltage calculations are fairly straightforward. Figuring the full drive current is at best a conservative guess because of sag. @Tom Kamphuys I liked your suggestion of just estimating the current off of the power output of the amp. 25W from 400V. That’s 63mA RMS current during full drive. If we add that on top of the 42mA DC current at idle, add additional 20mA RMS screen current in full drive, and then 13mA of preamp current we get

    63 mA (audio current)
    42 mA (dc power)
    20 mA (additional screen current)
    +13 mA (preamp current)
    ————
    138mA total during full drive. That’s seems to be in the ballpark of expectations. @elpico pointed out 150mA current rating should be plenty and it looks like the math bears that out.

    I think a lot of my confusion came from the concept of ripple current in the transformer and how it’s a lot higher than the current flowing out of the B+ node. I originally thought I had to include this when comparing the different transformer current ratings. I now have come to see that the rating that Hammond gives has an allowance for the higher ripple current and I can just use the B+ current (ballparked above) for selecting an appropriate sized transformer.

    @elpico I read that the AC15 was operated Class A. I don’t know what plate voltages and screen voltage it typically had. But based on the fact that I am pushing 25W of output power compared to its 18W, I have to assume I would need a bigger PT. Thank you for pointing out the similarities!
     
  18. XTRXTR

    XTRXTR TDPRI Member

    Posts:
    44
    Joined:
    Jul 13, 2020
    Location:
    Northern Colorado
    My two cents, It is nice to have room to add one or two more preamp tubes for mods later so the heater winding would need to be able to handle more current, and then you introduce a little more sag in the B+... Future tremolo, reverb, tube buffered FX loop.
     
IMPORTANT: Treat everyone here with respect, no matter how difficult!
No sex, drug, political, religion or hate discussion permitted here.