Do output transformers "saturate"

[trobbins]

Some background was that the test setup used a 100W Bass amp by Eminar with a Trimax TA 2418 110W output transformer with 1.7k PP primary, with each half winding measuring 22 ohm DCR, and 20H at 50Hz with 45V excitation. The B+ was direct from a beefy PT with doubler rectifier and 110uF, and the screen supply was from an LC off B+ comprising 6H with 240 ohm DCR and then 50uF, with 1k to each pentode screen. Output stage bias was fixed, and the measured 6L6 was a 'stronger' example as two of the 6L6 in the quad idled at a few mA less. Each 6L6 cathode had a 10 ohm sense resistor, so available B+ at peak cathode current point of 400mA was lowered by about 20V (assuming other 6L6 was only conducting about 300mA). Voltage probe loading on the anode was circa 13 Megohm, as I used a 12M VR68 to 1M fixed resistor divider to couple to my soundcard's 10Meg 10:1 probe.

Given Vgk was about 0V and Vak was close to zero at time of cathode current increasing from zero (due to open-circuit secondary), the screen current was likely quite sizeable as the cathode current rose up to the circa 100-150mA short plateau, which I'd guess would be the knee region that the valve was happy to be at for its given conditions. After about 3ms at that plateau, the cathode current then rises to about 400mA at which time I'd anticipate that the primary winding inductance collapsed and the anode voltage rapidly returned to B+. With the loadline point directly above B+ (minus circa 20V given the anode currents flowing), the cathode current ramps down by about 150mA over nearly 10ms, and the anode voltage perhaps slightly rises (due to less drop across the winding). With about 700mA load on B+, its voltage would sag about 60V over 10ms, depending on the timing of the 50Hz mains charging pulses.

The screen currents may all be down at circa 10-30mA when anodes are at nominal B+. The datasheet info is only at 300V screen, whereas screen supply voltage at start of test was likely circa 470V (assuming circa 30V drop across the choke) minus 1k drop of perhaps 10-30V. Certainly I could anticipate screen current to substantially rise when anode voltage is down near zero, but I don't see a significantly sagging screen voltage scenario at the time of 400mA sagging down to 250mA, as even the other side of 6L6's would have a nominal B+ level on their screens.

Maybe the impedance of the primary starts to rise and there is some coupling behavior between the half-windings. The core BH operating point could be starting to return toward the origin (ie. typical idle bias condition). The other primary half-winding's voltage doesn't appear to rapidly drop straight down to B+, and seems to show some droop during the time the cathode current on the other side is ramping down from 400mA.

All good fun trying to tease out the dynamic operating conditions

[Helmholtz]

Thanks for your reply.
I agree that plate voltage being close to B+ shouldn't cause excessive screen current.

Maybe a dynamic grid drive/ grid current effect?
When the loadline is essentially vertical, a little grid current might noticeably increase available plate current.

Also wondering why plate current is lower at the second cycle.

[Gregg]

So let's summarize and answer the topic question.
Under normal operating conditions guitar OTs DO NOT saturate.

[Greg_L]

So let's summarize and answer the topic question.
Under normal operating conditions guitar OTs DO NOT saturate.

That's way too simplified. You didn't even insult anyone. Gosh.

[The Dude]

Wow, what great and exciting discussions you had on the MEF in those days!
So sad I missed most of that.
Definitely should have joined way earlier, but was just to busy with my manager job.

Well, even if you were "late to the party", I'm sure glad to have you aboard. It didn't take you long to catch up with your wealth of knowledge.

[trobbins]

Also wondering why plate current is lower at the second cycle.

I'm thinking that may be related to dynamic driving conditions, given the 2 cycle signal and the 100nF coupling caps from the long-tail pair PI stage and possible blocking.

At the start of each signal cycle, the core BH operating point would transition over to the other 2 quadrants of operation, as the other primary half-winding then exhibits nominal inductance until it too reaches into saturation and the primary winding voltage collapses back to B+.

[Helmholtz]

At the start of each signal cycle, the core BH operating point would transition over to the other 2 quadrants of operation, as the other primary half-winding then exhibits nominal inductance until it too reaches into saturation and the primary winding voltage collapses back to B+.

In my words:

Half-primary voltage is (B+) - Vp. When Vp equals B+, primary voltage is zero.
By transformer principle the voltages at the 2 half-primaries should be mirror images no matter where the current flows.
Core operating point is determined by H which is proportional to total instaneous current, i.e. the sum of the winding currents.
So when winding voltage collapses on one side, it must at the same time also collapse on all other windings.

Zero primary voltage at non-zero current means zero impedance, i.e. full saturation.
The corresponding momentary loadline is vertical.

[Gregg]

That's way too simplified. You didn't even insult anyone. Gosh.

Life is complicated enough so for practical purposes it's as simple as that.

[Chuck H]

So let's summarize and answer the topic question.
Under normal operating conditions guitar OTs DO NOT saturate.

And to clarify (I think) "normal" in this case would include clipping the power tubes aggressively as is typical with many guitar amplifiers. The signal from any guitar amplifier preamp probably doesn't have what it takes to drive an OT into saturation before grid conduction equalizes limiting voltages.

As a non tech observer enjoying the analysis from my betters, am I interpreting this right?

[Chuck H]

Ok, so,..

To get to Mike K's interpretations of his listening experiences. He seems to want to equate the tonal performance of different OT's under heavily driven conditions by attributing it to at least partial, or maybe momentary saturation. Making some OT's sound buzzy while the MM iron performs better. Sort of reverse engineering the thread now, but what else might explain his sonic experience that seems to be OT specific?

I'm posting this because if OT saturation in guitar amps IS a myth then Mike will need other avenues to examine. In case anyone thought I wished badly for him. I don't. It would be great if he could have a real answer for his circumstances.

[trobbins]

The test I set up used a guitar amp, with the gain and vol controls maxed out on purpose to drive the output stage with effectively a squarewave. The 100W amp used a 110W rated OPT, so not noticeably undersized. The test condition used no load, although even with a nominal loadline the primary voltage swing would imho have been nearly the same for a similar V=L.di/dt and hence duration for the core to reach into saturation region. The test waveforms indicate that the core would have likely reached into the saturation region even with a 60-70Hz squarewave. The OPT was commercial but not 'hi-fi', and the cursory primary inductance measurements indicate the OPT was perhaps a reasonable bogey example for a guitar amp, and one could suggest that some guitar amps have OPT's with lower primary inductance, and hence are more prone to being pushed in to their saturation region.

Helmholtz - yes I see it the same way. The plots indicate the primary winding voltage collapses back to B+ (nominal low to zero volts across winding) for when the winding was in conduction, as well as when in cutoff. I can also comment that there is some leakage inductance energy exhibited (the inherent reason for the test) at the time of turn-off of the conducting half, which then forces the transient voltage overshoot on that winding (and hence the interest in the MOV management of that transient) as well as transfer of that leakage energy over to the other half-winding to force a transient voltage undershoot.

The other observation is that during cut-off, and after the other half-primary voltage has collapsed back to B+, the plots show the 'off' half-winding voltage noticeably changing (albeit to a minor level) and during the time that the other conducting winding has its current ramping down from a max level. To me that suggests the core BH operating point is starting to move back from its peak excursion level into saturation, and the loadline per se has some low impedance level.

I did take some X-Y plots during that test run, as per Loudthud's efforts, and observed quite reasonable plots clearly showing valve loadlines consistent with the loading resistance (or lack of), and also the vertical loading at B+. But I didn't capture a plot for the presented MOV related plots - perhaps as the persistence (even though I was using a digital scope) accumulates all excursions during the 2 cycle duration which makes for an arduous interpretation.

Ciao, Tim

[35L6]

What else ? How about primary impedance ? On the charts I looked at speaker impedance doubled at 5k . If the load line goes through the knee the load line for 5k would rotate counter clockwise and hit the shin at about Vgk equals -2 giving less gain . Same for bass resonance . This would result in a midrange emphasis . Does this make any sense ? I could be wrong , my understanding of magnatics is quite thin .

[Mike K]

Ok, so,..

To get to Mike K's interpretations of his listening experiences. He seems to want to equate the tonal performance of different OT's under heavily driven conditions by attributing it to at least partial, or maybe momentary saturation. Making some OT's sound buzzy while the MM iron performs better. Sort of reverse engineering the thread now, but what else might explain his sonic experience that seems to be OT specific?

I'm posting this because if OT saturation in guitar amps IS a myth then Mike will need other avenues to examine. In case anyone thought I wished badly for him. I don't. It would be great if he could have a real answer for his circumstances.

I don't expect anyone to know this.



To try to clarify what I perhaps inelegantly tried to propose in the first post: perhaps there's some reality here? I could guess, perhaps incorrectly, that Peavey's engineers know something, measured something and tried to replicate something they saw in a real amp. Perhaps not. But it seems to me they didn't come up with the idea out of thin air. I didn't mean to convey blame on anyone for it - I meant my question as to challenge what we know (what I don't know), and what this circuit was really trying to accomplish. But like so many things, sometimes reverse engineering doesn't really turn out great results. If you want to play that game, I'll give you some stuff and you can try to figure out what the heck I was thinking when I engineered something. I bet I can fool you...


And it's not just the OT - it's the whole circuit. Amps can be buzzy or fuzzy for a number of reasons. I'm not trying to imply or sell you on some specific iron. My case is not that scientific either. I'm not cloning amps; not usually. I typically am looking for transformers that can fill a certain niche, and often times I'm comparing slightly different designs, as we all are in cases of home brewing. I actually have very little interest in who winds the best "5E3" or whatever. Chances are I'm going to use that 5E3 iron on a different circuit and maybe I'll like the quirks of one vs another. So may you. If you want to build authentic vintage gear or make a current design sound best, that's just luck - or maybe it won't matter?

I know for sure some OTs, particularly small ones, are primary inductance limited and have a bandwidth limitation for their entire operation. For instance take a true 5C1 or AC4 OT - they are very small and have an enormous amount of "presence" built into them. To me, Fender and Vox did all their tone shaping with that OT and the speaker even though the VOX has a tone control, it's only a cut, not a presence like the tweed controls. Anyway, those are extremes but they show us how that happens. What I don't know is if they bandwidth limit as they are pushed. I know that any of them, no matter how good they are will degrade with too much signal, particularly lows or low-mids. If you put a treble booster out front you can push them more, like any amp, but those tiny cores will really start getting blurry. I'd guess this is maybe a precursor to true saturation. The sound is not good, the audio quality breaks down into what, in my mind, starts to sound like white noise. The bass may also start to drop off and perhaps the highs too. And this may happen in all transformers to some degree. Perhaps this is what we hear when we oversize cores. Empirically, I've done this and I've run 5W amps on everything from 5W cores to 50W cores, with the same impedance, and there's clearly a difference. On the ridiculously oversized core, no matter how you drive it with how much bass it seems to stay crystal clear. The harmonics are ear piercingly sharp and the volume noticeably higher. It's not always what you'd want, but when I recorded those vs others I was really impressed at how clear it came out in the recording. Same speaker and completely destroying it and it sounds as composed as recording a pedal through a big clean Fender. It's almost boring sounding.

I'm not sure where the cutoff is there when driving with a 5W amp, but I tend to use around a 12-18W sized core and to me, that has the clarity but it also doesn't sound sterile like the 50W (also it looks ridiculous having an OT that big on a tiny amp!). SE amps might be more sensitive to this kind of thing and I've never really done these kind of tests on PP amps. Actually, I've found I can get great tones out of 10W cores and 30W cores with the Class A PP amps, but I'm not really pushing even the 10W. But I do think there are probably noticeable differences in clarity and perhaps bandwidth changes. Because I'm not really trying to produce something other than something that sounds good to me, it's all a bit of trial and error and the best test is to plug it and and test it with as many speakers and guitars as you can. Not good for convincing anyone else or running an amp company that makes 1000's of units a year, but for someone building customs or to your own tastes, this is what you have.

[teemuk]

...could guess, perhaps incorrectly, that Peavey's engineers know something, measured something and tried to replicate something they saw in a real amp.

Maybe. But in the end that particular Peavey whitepaper topic would have more credibility if their "Saturation" circuit would have actually done what the whitepaper discusses to be the effect of OT saturation, that is *dynamic* limiting of the bandwidth.

But it doesn't. There is nothing dynamic in the Peavey circuit; the low and hi-pass corner frequencies are fixed to gain control setting and so far the idea isn't strikingly different from, say, driving amps with a Treble Booster pedal or using distortion effects or amps that have a prominent "treble boost" response before distorting stages baked in.

Hi-passing will eradicate quite a lot of low frequency IMD "flub" and "fart" (and is therefore a generally applied voicing method in many "high gain" circuits), low-passing will reduce IMD "fizz", but for an implementation in the year 1980 this sort of "voicing" idea is anything but proprietary or novel, and - more importantly- has practically nothing to do with output transformers.

Perhaps not. But it seems to me they didn't come up with the idea out of thin air.

Well, the actual idea Peavey used in the Saturation and Super Saturation generation of amps they might have just as well snatched from a 2203 JMP or just Tube Screamer or DS-1 pedal. Still far fetch to claim that the mid boost baked in response simulates "OT saturation".

That said, their tube emulation ideas for TransTube generation of designs are actually pretty sound and are really very, very good emulations of specific types of tube amps. I wish the whitepapers would have discussed this stuff deeper. Naturally all that "OT saturation" talk pretty much seizes when we jump to the TransTube designs, which I find a bit ironic too.

[Mike K]

Maybe. But in the end that particular Peavey whitepaper topic would have more credibility if their "Saturation" circuit would have actually done what the whitepaper discusses to be the effect of OT saturation, that is *dynamic* limiting of the bandwidth.

Again, the question is, are there any measurements that can show dynamic bandwidth does change BEFORE saturation i.e. due to effects of core loading, hysteresis, etc?

Hammond, for instance, shows a single power level frequency measurement for some of their transformers in the data sheet, but it's quite useless. They measure at an input signal level that is only a small fraction of what the core is rated for. The effect of OT on a clean amp is minimal unless it has limits as I mentioned above such as low primary inductance which limits bass response or high equivalent shunt capacitance. Those are quite easily heard, but they don't necessarily correlate to things that can be heard when the power amp is overdriven.