GU-50 Push-Pull AB OT Impedance/Load Line Questions

[defaced]

Got it. Thanks again for your help.

So getting to the power supply current requirements, it seems based on the shifted load line you drew in that the max current draw in AB1 is going to be around 360 ma. Taking LT's scope traces into consideration, it might ride up the 0v grid curve line till about 440 ma. So with those being max values, I'm thinking that specifying the PT to deliver 300 ma DC at the rated voltage (~570v AC) would be good. Does that sound about right?

[trobbins]

With respect to average power dissipation, one can interpret Loudthud's traces by comparing (integrating) the intensity of the trace when it is above max power dissipation line, and when trace is below. The cro screen persistence is a big aid to that assessment.

[defaced]

That makes sense. I was interpreting the plots too literally. There's a time variable that's not explicitly shown, but can be interpreted by looking at the brightness of the lines on the CRT.

[trobbins]

Yes the persistence represents the relative time spent at any point in the V-I locus.

In Loudthud's resistive load trace, the loadline is somewhat the same intensity along the line, with a drop off in intensity corresponding to the peak regions of a sine-wave, and an increase in intensity in general in the cut-off region due to the relative compression of the 'length' of that part of the loadline. So that plot is relatively easy to interpret with respect to how much time the anode spends dissipating a higher power level (ie. probably only above the 'max design dissipation' level for 15-30% duty-cycle).

Looking at the inductive loadline trace, one can assume that 50% of time is in cutoff region (ie. horizontal region to right of bias point). So your 'power dissipation' sketch could be perhaps made a little more accurate with respect to dissipation amplitude and timing.

[loudthud]

A few words on how the pics were created.

The scope is a Tektronix 7104 setup for X-Y display using two 7A26 plugin amplifiers. A high voltage 1000X probe connected to the plate is connected to the horizontal input. A 10X probe connected to a 1 ohm resistor in the cathode is hooked to the vertical input. Signal source is a Gibson SG. The first shot with the high dissipation was just a transient bump of the strings, not any note or cord. I just wanted to show that clipped waveforms into reactive load are a long way from resistive load lines. In solid state amps it's critical.

The oscilloscope display is recorded with an old Sony camera that can shoot a low resolution 5 second video clip. The video clips are viewed by Windows Media Player and paused at interesting places. Because of video compression, you can't pause and look at every frame as you would like. Also, since the guitar signal is not synchronous with anything, you can't read too much into the brightness seen in any video frame. A dim line might just be fading out and a bright line more recent when captured by the camera.

To judge actual plate dissipation you would need an analog multiplyer to produce a waveform of instantaneous dissipation and then an averaging circuit or in a digital domain, digitize current and voltage and process that data.

[defaced]

Very cool setup LT. Thanks for sharing. Those shots explained more to me about how a power stage actually works than anything I've come across to date.

I'm looking at stuff I have laying around so I can mock up a power stage to play with, and the only OT I have that could stretch is a Hammond 1650T, which has a P-P impedance of 1.9k. So if I mismatch it 4:1 (16R load on 4R tap), I can get close enough to the 8K to mock up a stage and see how it runs. I know 2:1 is often done with OTs, but I haven't seen anyone do 4:1. Is this doable?

[es345]

I think it shoul be ok for a test. I assume that the isolation inside the OT is good enough for 3 KV to have some safety margin. I don't know the detailed spec. You might see a degradation at full power in the lower frequencies.

One comment regarding the power supply: I would choose a 3 rail config with the first rail resulting in about 310V DC screen supply and second/third rail stacked above resulting in 850V DC plate supply when idle. These voltages will go down to about 300V and 800V at full power. This will allow to use cheaper filter caps.
Important: the screen supply should be filtered with CLC, otherwise you will notice a significant 100 or 120 Hz screen modulation when increasing the output power. Have a look into the power supply I have in my bass amp.

[defaced]

I will see what I can find out regarding the voltage rating of the OT insulation. That was my primary concern too.

I had read that the screens needed to be well filtered, so your CLC recommendation makes perfect sense. I have not settled on a PSU design yet (I need to work out the PI and buffer circuit first - both will be MOSFET), but had considered regulating the screens with a high voltage MOSFET circuit. The current plan is to have the PI, buffers, FX loop and preamp circuit start off with a 400v node. I planned on making that node using a MOSFET regulator, so adding another one isn't a big deal.

[defaced]

I think it shoul be ok for a test. I assume that the isolation inside the OT is good enough for 3 KV to have some safety margin. I don't know the detailed spec. You might see a degradation at full power in the lower frequencies.

One comment regarding the power supply: I would choose a 3 rail config with the first rail resulting in about 310V DC screen supply and second/third rail stacked above resulting in 850V DC plate supply when idle. These voltages will go down to about 300V and 800V at full power. This will allow to use cheaper filter caps.
Important: the screen supply should be filtered with CLC, otherwise you will notice a significant 100 or 120 Hz screen modulation when increasing the output power. Have a look into the power supply I have in my bass amp.

I got ahold of Hoffman and they hi-pot test their OTs at 2500v for a short duration. The gent I'm talking to says a max working voltage might be around 1KV RMS, so 1400v peak. From what I'm reading, during cut off, plate voltage can rise to twice the B+, so 1600v, which means that that this OT is not not really a safe option for any high powered tests. Which brings me to my next question, did you have an OT wound for your bass amp, or did you use an OT you already had?

[es345]

It's a customized toroid OT according to my requirements, manufactured by a local supplier here in Germany/Poland.

main requirements:
- designed for 350W
- frequency response from 20 Hz to 20kHz at full power 350W (-3 db limit)
- Raa 2kOhm, Rcu anode to center 50 Ohm, designed for Ub=950 at idle Mode
- secondary 4 Ohm

The result:
Rcu anode to center 67 Ohm, a little bit higher than planned.
On the other side the frequency response shows 400W down to 20 Hz, better than expected.

[defaced]

Ok, that's about what I expected. Did you have any problems finding a supplier that could meet the voltage requirement?

[tedmich]

if you need 1600V peak and 1.13kV RMS you are probably looking at some spendy iron, Lundahl spendy...

Kind of spoils the cheap tubes IMHO

[defaced]

In the short term, yes, but shopping and quoting are free, so I'll see what falls out. If I have to parallel them to get the power I want and drop the B+, I will; I already have four sockets and plenty of tubes. This is as much an academic exercise in how to design it as it a desire to build a real amp.

[tedmich]

This is as much an academic exercise in how to design it as it a desire to build a real amp.

as long as there are no final grades, i say go for it!

[es345]

Did you have any problems finding a supplier that could meet the voltage requirement?

It took a while to find a supplier for customized toroid OT's. Isolation voltage wasn't a problem. The challenge in my case was to meet output power,copper resisistance and lower frequency on one side and weight plus size limits on the other size. It took us 3 iterations for the best compromise. That was expensive for sure but for personally for me it was a little bit of research and fun.

If you go for EI core without limitations regarding wight and size it is much easier. I have good experience with this company RITTER electronic systems--Die Röhrenspezialisten seit 30 Jahren!, the homepage is written in German. They have an OT for GU50 PP with Raa 8K as a standard product in their HIFI PP section.