Using mains transformer as output transformer

[Elias]

Hello,

Interesting. So many good experience with toroids. Sounds very promising for me.

326a:
2mA tolerance comes to +/-5% which can be hard to maintain during long periods if considering tube ageing? Or can we assume tube bias current will allways stay within +/-5%?

Maybe can use a correction circuit for setting the toroid secondary DC bias. The gain in the DC feedforward path is the same as toroid winding ratio. This should maintain correct compensation current in the secondary regardless of tube bias current changes. Can even tune the tube bias if you wish to try how it affects the sound and at the same time keep the toroid in it's optimum performance. I'm sure this idea is not new even I don't recall seeing it before.

- Elias

[ErikdeBest]

Hi Elias

You are really good in reinventing the wheel! There is a quite recent thread were a member presents an amplifier with exactly this topology, a tracking CCS in one of the windings.

Another kind of hybrid

Yesterday I looked into my toroidal stock and found some 230V to 24V + 24V units. I thought about using one for output transformer for a 6C33C operated at about 250mA. Soon I concluded that I would need to inject about 2,5A in the 24V winding. I measured the DC resistance of the 24V winding which should be about 0.3R...injecting 2,5A would develop 1V, which is not much, but then, how to accomplish a fine control over the 2,5A? I thought about using an LM338 set as CCS for the job, but there would be at least a 4V drop over it and the current setting resistor...that is 10W dissipation. Hmmm...better would be to use a trafo with 2 110V windings, one for the output tube and the other for DC cancellation...but the 230V winding is much more appealing, as it will present larger inductance.

Maybe a toroidal with dual 18V secondaries. This one would need about 3.2A in the secondary to cancel 250mA in the primary. One could then feed the 6C33C with 12VDC/3.3A, put one of the 18V secondary in line with the cathode and set the 6C33C bias for best cancellation.

just some thoughts...

Erik

[316a]

[2mA tolerance comes to +/-5% which can be hard to maintain during long periods if considering tube ageing? Or can we assume tube bias current will allways stay within +/-5%?

Maybe can use a correction circuit for setting the toroid secondary DC bias. The gain in the DC feedforward path is the same as toroid winding ratio. This should maintain correct compensation current in the secondary regardless of tube bias current changes. Can even tune the tube bias if you wish to try how it affects the sound and at the same time keep the toroid in it's optimum performance. I'm sure this idea is not new even I don't recall seeing it before.

The 2mA tolerance either side of 40mA seems to have no discernable effect , after that it takes much more offset for any noticeable distortion to creep in . Best to keep it simple rather than add feedforward or whatever circuitry ! Much better off hooking up a sig gen and DVM every now and again and use variable grid bias to adjust . It's a tweaky circuit , bypassed CCS in the cathode circuit can help but then again the DC filament supply needs to be regulated too . Never tried this as I used a choke input filament supply .

At the end of the day there is a price to pay for using a cheap transformer , as the offset choke and DC filament supply will increase the cost . It's a fun experiment if you have the parts but I'd recommend a pair of Trancendar 3K SE transformers instead . For the price and performance of these units there's no point messing around with chokes and mains transformers !

cheers

316a

[Elias]

Hello,

Erik:
Thanks for the compliment

No that is a one long thread! I need to take some time to go it through.

I would like to have a toroid with for example 230V / 12V + 110V windings. Could reduce the compensation current quite a bit.


316a:
I see no problem putting a feedforward DC feed. Basically all you need for it is one opamp and one power transistor for current generator + some small signal stuff. The price for these parts almost nothing.
Why do you need the choke for? If you put one secondary winding in series with filaments, and drive it with constant current, there is no need for the choke. By definition constant current source will keep the filament in constant potential also. There is no way the signal can be fed back to filaments.

Well, buying a ready made output transformer kills the DIY in it


For the passive DC current tracking, can use simple metering circuit. The ratio of R1 and R2 is the same as toroid winding ratio.

I think I need to put the meter in the front panel of the amp, because otherwise it makes me worried if the bias is correct and it prevents me from enjoying the music


- Elias

[316a]

Originally posted by Elias
316a:
I see no problem putting a feedforward DC feed. Basically all you need for it is one opamp and one power transistor for current generator + some small signal stuff. The price for these parts almost nothing.
Why do you need the choke for? If you put one secondary winding in series with filaments, and drive it with constant current, there is no need for the choke. By definition constant current source will keep the filament in constant potential also. There is no way the signal can be fed back to filaments.

Well, buying a ready made output transformer kills the DIY in it


For the passive DC current tracking, can use simple metering circuit. The ratio of R1 and R2 is the same as toroid winding ratio.

I think I need to put the meter in the front panel of the amp, because otherwise it makes me worried if the bias is correct and it prevents me from enjoying the music


- Elias

I think you're looking far too deeply into this , maybe perhaps you should build one of these amps and report back ? In the case of a 6B4G , a one amp CCS will not only get hot running with enough compliance but also may have stability problems . You cannot use a centre reading meter with a directly heated valve unless I'm mistaken . I do not see the point in heating an indirectly heated power valve with DC as in your diagram , not only does it add complexity but also cost , which negates the whole point of using a mains transformer in the first place !

cheers

316a

[Shoog]

I have messed about with toroidals as outputs a lot. I tried choke loaded SE outputs and Valve CCS loaded outputs. Choke is the better way to go because you lose less voltage. However I think the whole idea of SE toroidals is a big mistake and wouldn't recommend it at all. Having said that, DC compensation would seem to be the way to go. If I did that I would put a CCSink in the cathode to make it rock solid stable over time. Basically choke loads are expensive and introduce all of the nasties which toroidals avoid - so no advantage and no cost saving.

I have three amps using toroidals in PP amps and this is where they shine the most. Response down to 10Hz and up to 60kHz. DC balance is essential and can be achieved with CCsinks in the cathodes, or with "Garter Bias". I like to get my CCS to within 1mA, bass suffers badly after that. LM317's make reasonable candidates for CCS. One thing to watch is because of their extreme bandwidth they do tend to ring and may need dummy loads. I have found that they consistently perform better with the primary winding in one direction against the other. This is a interwinding capacitance issue which both kills the bottom end and produces a rising response centered around 50khz. Reverse the windings and this problem goes away.
Used as phase splitters they benefit from a bit of a step down.

Shoog

[Michael Koster]

I thought DC balance would work on the toroid just fine,
as long as there are 2 complementary devices to cancel the
average current, i.e. push-pull. In SE I have only used parafeed
with a capacitor.

The circuit at
Another kind of hybrid
uses a MOSFET opposite a tube with a PP OPT. The MOSFET is
controlled to both balance DC current and actively drive signal
current inverse to the tube's signal current. I also experimented
with just DC cancellation.

I have noticed some interesting transformer saturation effects
based on different DC balance schemes.

Basically, what happens is that if you balance the DC at idle,
then at full signal there is a DC shift due to the load-line
asymmetry. The increasing current swing is greater than the
decreasing current swing, and when a large signal is applied
the average current increases. If the DC balancer is fixed and
adjusted at idle, there will be a DC offset at large signal.

The PP transformer I used tolerated a few MA of unbalanced
DC current, so I didn't have a lot of saturation problems to
deal with. Your experiences with the toroid will be interesting.

Michael

PS The "zero crossing meter" if damped properly would display
this DC shift with signal

[Merlinb]

Originally posted by Elias
Hello,

How about bandwidth? I would guess that because the big and heavy toroid is initially designed for 50Hz, it could support low frequencies quite ok? I don't know what will happen at 20kHz though?

That depends mainly on the primary inductance. If you want any bass at all you should be looking for the highest primary inductance you can get.
Cheap toroids may have less than 1 Henry, which is useless. You might get lucky and find one with around 10 Henrys, which would be "useable" for lo-fi applications using pentodes (such as a guitar amp or radio), but if you want to get anything that remotely looks like hifi you''ll have to use it with a low impedance valve, like a nice beefy triode. Couple of triode-strapped ECL86s maybe.

[Shoog]

My personal experience of using toroidals is with
- 6080's running at 100V 100ma, very low output impedance.
- ECL82 with plate to plate feedback which reduces output impedance down to CF levels.
- 807 with plate to plate feedback as well.

So all of my applications have had very low output impedance finals which makes my experience less than generally applicable.

That depends mainly on the primary inductance. If you want any bass at all you should be looking for the highest primary inductance you can get.
Cheap toroids may have less than 1 Henry, which is useless. You might get lucky and find one with around 10 Henrys, which would be "useable" for lo-fi applications using pentodes (such as a guitar amp or radio), but if you want to get anything that remotely looks like hifi you''ll have to use it with a low impedance valve, like a nice beefy triode. Couple of triode-strapped ECL86s maybe.

However I believe Steve Bench made his Matrix amps with toroidals as outputs and he measured very high inductances and very good results. I think experience is stacking up to show that they work well and work well down to very low frequencies - better than many SE transformers.

Shoog

[Elias]

Hello,

Ok, I did some measurements. I had 115V+115V / 43V / 105VA toroid. One primary I fed with signal generator with 100ohm impedance, the other primary I fed with 0...100mA DC current with 470ohm series resistor and the secondary I loaded with 8ohm resistor. I monitored sinusoidal voltage over the load resistor with oscilloscope. The signal power transferred to the load was about 1W.

The result:
I observe no change in waveform when I change the DC current. No change at all! Waveform at the load remain exactly the same with or without DC current in primary.

My conclusion:
I have no explanation other than this toroid does not saturate with 100mA at primary coil.


Can it be because the DC level I used is well within the power margin of this 105VA toroid??

Or is this toroid made of some extraordinary good material??


All I know about this toroid is the type number Salcomp FM1002 and it's industry surplus I bought for less than 2€ and I have a box of them


If my conclusion if valid, I should have no problems using this in SE output.

Or in the other words, what am I doing wrong?


- Elias