OT, the bigger the better? And which one to choose?

[Steve Conner]

Do a google search for "Hartley peavey white papers". You might as well read them all.

You're confusing load current with magnetising current. The excursions into Class-AB ought to be symmetrical, they'll only contribute to saturation if one side of the PP output stage develops a different volt-second product to the other one.

[Alan0354]

Do a google search for "Hartley peavey white papers". You might as well read them all.

You're confusing load current with magnetising current. The excursions into Class-AB ought to be symmetrical, they'll only contribute to saturation if one side of the PP output stage develops a different volt-second product to the other one.

Thanks

I don't quite get this, I am not very good with magnetics, so bare with me. My understanding is the current line up the cluster of molecules in the metal, and when it all line up, then it get into the saturation. That current can be from the AC signal or DC offset current. As long as the two side of the push pull is not balance, there is an offset current. In class AB, there are part of the cycle one side is off, while the other side keep increasing the current. There is an offset current developed. Only the class A that both side are on 100% of the time that the net current is zero.

Please explain to me, thanks for your time first.

I have been avoiding magnetics and transformers all these years, we designed a lot of HV switching power supplies into 10KV and beyond. I always took the short cut by having my engineer designed the transformer and the switching circuit. I only got involved in designing the closed loop stability control part. Who can think one day ( like now) that I got into guitar amp, that I so wish I had involved in the design. I want to use off line switchers for guitar amp, I could have done it if I were to have more involvement with the HV switchers before. We designed really robust supply. We never have burning issue!!!



Alan

[Steve Conner]

Yes, "the current line up the molecules of the metal" but it's not the current supplied by either of the tubes, or the current delivered to the speaker, that does it. The classical model for a transformer includes an inductor shunted across one of the windings of the "ideal transformer", and it is the current in this inductor that drives the core flux.

Because the inductor is in parallel with the load, the load current doesn't affect the current in it.

And because the model transformer is "ideal", it transforms at any frequency. The magnetising inductance is linked to all of the windings: it's the sum of the amp-turns in all the windings that drives it. Hence why a push-pull OT doesn't saturate with DC: the sum of amp-turns is zero.

Finally, the current in the magnetising inductor looks like the integral of the voltage waveform. This is why we talk of volt-seconds. Too many volt-seconds is what saturates a transformer, not too much current.

[Alan0354]

Yes, "the current line up the molecules of the metal" but it's not the current supplied by either of the tubes, or the current delivered to the speaker, that does it. The classical model for a transformer includes an inductor shunted across one of the windings of the "ideal transformer", and it is the current in this inductor that drives the core flux.

Because the inductor is in parallel with the load, the load current doesn't affect the current in it.

And because the model transformer is "ideal", it transforms at any frequency. The magnetising inductance is linked to all of the windings: it's the sum of the amp-turns in all the windings that drives it. Hence why a push-pull OT doesn't saturate with DC: the sum of amp-turns is zero.

Finally, the current in the magnetising inductor looks like the integral of the voltage waveform. This is why we talk of volt-seconds. Too many volt-seconds is what saturates a transformer, not too much current.

Thanks

This is over my head!!!! I really need to read up this. I have been avoiding transformers, and it is not that hard to totally avoid transformer in the EE field!!!

So you are saying if I want more compression, I pull DC current from one side of the primary winding?

[Steve Conner]

Well, I don't know how you could have avoided magnetics if you got involved in switching power supplies. For instance, the inductance of the filter chokes varies with the load current, and you have the boundary between discontinuous and continuous conduction, and both of those impact your control loop.

Yes, pulling DC through a winding would cause compression. But what I'm saying is more subtle, I believe that many classic tube amps accidentally did this when overdriven, and the compression is part of their sound.

Some computer scientist coined the phrase "Worse Is Better" to explain how Microsoft products got so popular. I believe classic guitar amps are the same. The circuitry is wrong in ways that would make a rocket scientist EE poke his eyes out with the soldering iron, or more likely have his technician do it for him to keep the trade union happy. But the mistakes are the very thing that make the amps sound good! To understand good sound you have to study bad engineering.

[Alan0354]

Well, I don't know how you could have avoided magnetics if you got involved in switching power supplies. For instance, the inductance of the filter chokes varies with the load current, and you have the boundary between discontinuous and continuous conduction, and both of those impact your control loop.

Yes, pulling DC through a winding would cause compression. But what I'm saying is more subtle, I believe that many classic tube amps accidentally did this when overdriven, and the compression is part of their sound.

Some computer scientist coined the phrase "Worse Is Better" to explain how Microsoft products got so popular. I believe classic guitar amps are the same. The circuitry is wrong in ways that would make a rocket scientist EE poke his eyes out with the soldering iron, or more likely have his technician do it for him to keep the trade union happy. But the mistakes are the very thing that make the amps sound good! To understand good sound you have to study bad engineering.

Ha ha, I updated my post #62 while you replied at the same time. I explained how I weasel out of designing the transformer and kick it to the engineer that work for me, and I just did the closed loop stability design. I was the manager, I could switch things to my liking and I was very into MOSFET high voltage pulsers and RF stuff. Yes, I regret every bit of it, I never really thought one day I would really getting back into this, I though this chapter had long closed. I had the switching supply in mind if one day I do get back into guitar amp, I even got a board from my engineer in the late 90s that I can modify. But I also had a fall out with the guy as he is not very good other than in switching power supply and that's not good enough for me. So I am left cold now!!!

And yes, you can do the characterization by treating it like a black box for poles and zeros. I don't need to know how the transformer work or even the speaker, all I care is what is the phase shift from input and output and what is the amplitude change. I have not done the calculation and taming of the tube amp, but combination of OT and speaker can be treated as a black box. I think you just wear ear plugs while you doing the characterization and apologize to your neighbor!!!

[Steve Conner]

The black box approach only tells you about the exact black box you have on your bench today, at one signal amplitude. If you want to design a control loop for a nonlinear black box that you intend to mass-produce, then you have to know what's inside so you can work out what will be the worst-case combination of component tolerances, temperature and signal level. It would take far too long to explore the whole parameter space blindly. Then if you like Laplace transforms, you can do them on a linear approximation to this worst-case system.

Another way of looking at conditional stability is that the designer failed to identify the worst-case system. Of course reality will be kind enough to find it for him in production.

[jpfamps]

As far as I know, they were all either straight-wound or once interleaved: the cheapest and second cheapest options respectively. If I didn't know better, I'd say that the amps with feedback would be once interleaved, and those without feedback (AC30, 5E3, 5F1, etc) would be straight wound. However, I could be wrong, I believe jpfamps or someone has a recipe for a Vox AC30 transformer.

The JMI era Woden transformer AC30 is 5 sections: 1/4 primary, 1/2 secondary, 1/2 primary, 1/2 secondary, 1/4 primary. The two outside section make one half of the primary and the middle section make the other half of the primary. This way you can get the two halves of the primary well balance for all parameters only using one wire gauge for the primary, which simplifies the winding procedure.

From reading around the subject, this pattern seems to be regard as the most "bang for buck".

Logically, I might have expected a transformer without negative feedback to have more sections as the frequency response of the amp is not extended by negative feedback, however with guitar amp design then logic is often sacrificed at the altar of economics.

The Tiny Terror uses a similar 5 section design, as do the Drake transformers used in Marshalls from the late 60s onwards.

The Twin Reverb also uses a 5 section transformer, but the interleaving pattern is slightly different: 1/4 secondary, 1/2 primary, 1/2 secondary, 1/2 primary, 1/4 secondary. I think this arrangement was used as the transformer still benefits from having a 1/2 section on the outside, and as Fender don't tap the secondary winding, there is no issue with utilization of the transformer properly.

Incidentally, the JMI-era AC30, Marshall 50W and Fender Twin all use the same sized core for the transformer.

Partridge transformers used in Hiwatts etc, have a more complex interleaving arrangement. John Chambers from Champ Electronics did explain this to me a few years ago over the phone, but I have failed to remember the exact details. I do remember that the 50 and 100W version have slightly different arrangements.

[Alan0354]

The black box approach only tells you about the exact black box you have on your bench today, at one signal amplitude. If you want to design a control loop for a nonlinear black box that you intend to mass-produce, then you have to know what's inside so you can work out what will be the worst-case combination of component tolerances, temperature and signal level. It would take far too long to explore the whole parameter space blindly. Then if you like Laplace transforms, you can do them on a linear approximation to this worst-case system.

Another way of looking at conditional stability is that the designer failed to identify the worst-case system. Of course reality will be kind enough to find it for him in production.

From my experience, it is the small signal that is really important for stability control. In fact, most of the time I use very small signal to characterization.

I did not do the guitar amp yet as the discussion was mainly about the use of Laplace Transform and Bode Plot. The little experience that I have with guitar power amp, the open loop gain is so low, the poles are usually so high that you pretty much use dominant pole compensation like Fender putting a 2200pF on each leg at the plate of the PI and called it a day. I never see anything unstable about it. So I really never get into this part. It was just mentioned that the interleaving OT provide more phase margin that get me started the whole thing here.

Using black box approach worked for me. Remember in feedback theory, it's all about at the 0db crossover. Basically you disconnect the feed back divider to the negative input of the amplifier ( the other side of the PI in this case). Ultimately You feed the signal at the positive input and look at the signal at the the voltage divider. You want to look for the phase shift at the voltage divider to be less that 180 deg when the amplitude roll off the the same amplitude as the input. That's the 0db cross over. And if the phase shift has 45 deg margin, you get a stable system. I have not worked on the clipping amp, but a lot of the system or power supply I worked with don't have a linear sine wave response either, my experience is you do the small signal design, I never see any problem yet. But of cause you chop the whole circuit into smaller section to characterize them separately so the gain is not as high at low frequency.

You do verify by pulsing and look at the over shoot and ringing. After you design the system, you test by injecting a pulse and look at how the loop response. Also you turn on and off the power and get it to oscillation. When checking for stability, it's kind of have to have a nose for it also. I looked enough of these you can kind of tell whether the loop is stable or at the verge of busting into oscillation. I was from the days that we design discrete op-amp and sample-and-hold amp in Lecroy of like a 200MHz. Taming those amps are a big job, you really get a feel of how a stable system feels compare to some that are very twitchy...........I think twitchy is really the best word to describe!!!

[Enzo]

Her, all in one place, Mr Peavey's white papers and a bunch of other stuff.

[Alan0354]

The JMI era Woden transformer AC30 is 5 sections: 1/4 primary, 1/2 secondary, 1/2 primary, 1/2 secondary, 1/4 primary. The two outside section make one half of the primary and the middle section make the other half of the primary. This way you can get the two halves of the primary well balance for all parameters only using one wire gauge for the primary, which simplifies the winding procedure.

From reading around the subject, this pattern seems to be regard as the most "bang for buck".

Logically, I might have expected a transformer without negative feedback to have more sections as the frequency response of the amp is not extended by negative feedback, however with guitar amp design then logic is often sacrificed at the altar of economics.

The Tiny Terror uses a similar 5 section design, as do the Drake transformers used in Marshalls from the late 60s onwards.

The Twin Reverb also uses a 5 section transformer, but the interleaving pattern is slightly different: 1/4 secondary, 1/2 primary, 1/2 secondary, 1/2 primary, 1/4 secondary. I think this arrangement was used as the transformer still benefits from having a 1/2 section on the outside, and as Fender don't tap the secondary winding, there is no issue with utilization of the transformer properly.

Incidentally, the JMI-era AC30, Marshall 50W and Fender Twin all use the same sized core for the transformer.

Partridge transformers used in Hiwatts etc, have a more complex interleaving arrangement. John Chambers from Champ Electronics did explain this to me a few years ago over the phone, but I have failed to remember the exact details. I do remember that the 50 and 100W version have slightly different arrangements.

You make me very curious to open the Magnetic Components OT that is supposed to be made to spec for Fender Vibrolux/Pro and Bandmaster to see whether it is interleaving or not. Bottom line, I really don't care as long as it is $46 and there are quite a few making OT for Vibrolux/Bandmaster for under $50, Harmony included. So as long as it works, I am happy camper. As I said, to me, it's about the money and I am allergic to expensive stuff.

[Alan0354]

Her, all in one place, Mr Peavey's white papers and a bunch of other stuff.

Where?

[frus]

As far as I know, they were all either straight-wound or once interleaved: the cheapest and second cheapest options respectively. Compression can be thought of as a dynamic limiting of the bandwidth. As the transformer core saturates (and as I pointed out above, this only happens from DC imbalance, not loud signals) its magnetic permeability decreases. This lowers the transformer's magnetising inductance, causing the bass end to be high-pass filtered. I think it also increases the leakage inductance, low-pass filtering the treble. This effect would be lessened by interleaving since there is less leakage inductance to start with.

but surely, you need more iron even in P-P transformer for more power? or is it some other mechanism (not saturation)?
thanks for your explanations, they are very illuminating!!!

[Steve Conner]

Sorry, I should have said: If the transformer is properly designed with enough iron to pass the lowest note on the guitar at full power without saturating (and I believe this was true for classic OTs) then saturation can only be caused by DC imbalance.

[Alan0354]

I have a question, so the signal power don't cause saturation, only the offset DC current cause saturation. Then why is it that the small core transformer saturate at low frequency? Better yet, what is limiting a 15W OT from using in a 50W amp?