For valve power supplies I have been using back to back transformers. Drop mains down to 6vac for heaters then transform it back up to a round 200VAC using transformer backwards. This seems to have worked well.
I was wondering if I can do a similar thing with output transformers. Output transformers are very expensive and sometimes hard to find. The ratio needed from primary to secondary is about 1000:1. So output of valve amp goes into a 240VAC:12VAc transformer then into a another 240VAC:12VAC transformer.
Would this work ?
I was wondering if I can do a similar thing with output transformers. Output transformers are very expensive and sometimes hard to find. The ratio needed from primary to secondary is about 1000:1. So output of valve amp goes into a 240VAC:12VAc transformer then into a another 240VAC:12VAC transformer.
Would this work ?
Last edited:
A 20:1 voltage turns ratio using throw-away 50W qu-hal downlight transformers with 0-120-240V primaries has been used by a few amps for an output transformer to give 3.2kohm PP to 8 ohm speaker matching.
Given that many of those downlight transformers have only a 0-240V primary, they have been mostly used for SE output transformers up to about 10W with great results, and also just simply as back-to-back transformers for the power transformer role to give both 12V heater and B+ levels.
Given that many of those downlight transformers have only a 0-240V primary, they have been mostly used for SE output transformers up to about 10W with great results, and also just simply as back-to-back transformers for the power transformer role to give both 12V heater and B+ levels.
I was a bit out with my 8000:8 giving a 1000 times ratio. I will go back and read up on transformer theory and work out impedance ratios.
Nigel, the thing to know about transformers and turns-ratios is that the impedance ratio is the square of the turns ratio. The voltage step up (or down) is proportionate to the turns ratio. The matching to the amplifier, to the impedance ratio.
8,000 : 8 Ω impedance ratio is √(x/y) = √( 8000 ÷ 8 ) = 31.6 turns ratio. In 0–120–240 primary land, that'd be 120 ÷ 31.6 → 3.8 V secondary. In the 240 primary land, the secondary would be 7.5 volts. Kind of an unusual transformer.
But then 8,000 Ω is kind of unusual for a modest duty output stage. Lots of volts, little amps. SOME OTHER THINGS…
_______
Another would be the field saturation limit of the repurposed mains-as-output transformer. Power transformers are generally designed to be quite near core saturation running at 50 Hz, since no commercial power is below that frequency worldwide. What that means is that the core is near (and usually even slightly over!) saturation when 220 V of signal RMS is on the primary side, at 50 Hz. At 30 Hz, the saturation voltage would be ³⁰/₅₀ × 220 = 132 volts of RMS swing. At 20 Hz, 88 volts. If you're using a power transformer with a 9 volt secondary and a 220 V primary (²²⁰/₉ = 24:1 … squared ≈ 600×:x impedance matching), remember the derating for lower frequencies.
Another poster (whose name I'm forgetting) is an advocate of using TRIAD power transformers of the circular-ring core type. Toroidal. Claims excellent application and sonic fidelity.
_______
Then there is “iron lamination thickness”. Again, your power line transformer is working at top of 60 Hz. Since it costs more to interleave more, thinner lamination sheets than fewer thicker ones, the power transformer makers are inclined to waste a little power of the transformer to eddy current heating in the core. This is at 60 Hz. The amount of ohmic heating goes up linearly with the frequency, and goes down inverse-square with the lamination thickness. Is your chosen power transformer really up to passing a lot of 5-to–20 kHz signal should the music require it?
Just saying.
GoatGuy
8,000 : 8 Ω impedance ratio is √(x/y) = √( 8000 ÷ 8 ) = 31.6 turns ratio. In 0–120–240 primary land, that'd be 120 ÷ 31.6 → 3.8 V secondary. In the 240 primary land, the secondary would be 7.5 volts. Kind of an unusual transformer.
But then 8,000 Ω is kind of unusual for a modest duty output stage. Lots of volts, little amps. SOME OTHER THINGS…
_______
Another would be the field saturation limit of the repurposed mains-as-output transformer. Power transformers are generally designed to be quite near core saturation running at 50 Hz, since no commercial power is below that frequency worldwide. What that means is that the core is near (and usually even slightly over!) saturation when 220 V of signal RMS is on the primary side, at 50 Hz. At 30 Hz, the saturation voltage would be ³⁰/₅₀ × 220 = 132 volts of RMS swing. At 20 Hz, 88 volts. If you're using a power transformer with a 9 volt secondary and a 220 V primary (²²⁰/₉ = 24:1 … squared ≈ 600×:x impedance matching), remember the derating for lower frequencies.
Another poster (whose name I'm forgetting) is an advocate of using TRIAD power transformers of the circular-ring core type. Toroidal. Claims excellent application and sonic fidelity.
_______
Then there is “iron lamination thickness”. Again, your power line transformer is working at top of 60 Hz. Since it costs more to interleave more, thinner lamination sheets than fewer thicker ones, the power transformer makers are inclined to waste a little power of the transformer to eddy current heating in the core. This is at 60 Hz. The amount of ohmic heating goes up linearly with the frequency, and goes down inverse-square with the lamination thickness. Is your chosen power transformer really up to passing a lot of 5-to–20 kHz signal should the music require it?
Just saying.
GoatGuy
Thanks for the detailed explanation. Clearly the power supply side function is fine because it is 50Hz.
I did look up the transformer impedance ratios and found the mains to 12 volt transformer to be 3200 ohms so it is about right for my amp. However as you say heating will be a problem at higher frequencies.
I had a similar problem on a class d amp I designed, the output inductor got to 120 degrees C ! I was using the wrong core and it needed a t106-2 and not a power inductor core.
I did look up the transformer impedance ratios and found the mains to 12 volt transformer to be 3200 ohms so it is about right for my amp. However as you say heating will be a problem at higher frequencies.
I had a similar problem on a class d amp I designed, the output inductor got to 120 degrees C ! I was using the wrong core and it needed a t106-2 and not a power inductor core.
They have found good use for output transformers in guitar amps - so an excellent diy recycling application. One OPT comparison by Grant Wills: Background to a new SE amplifier
You *need* a gap for SE use. Period.
Even the cheesiest cheapest transformers used in the cheesiest cheapest radios in the 40´s 50´s knew so and didn´t try to skimp on *that* .
At the very least, they loaded all E one way and closed the magnetic circuit with all I stacked together, trusting natural punching irregularities to provide a built in gap.
next step up was to add a sheet of paper (newsprint is fine, better than nothing) to guarantee a minimum gap.
Just think about the "savings" in not doing so :O
Even the cheesiest cheapest transformers used in the cheesiest cheapest radios in the 40´s 50´s knew so and didn´t try to skimp on *that* .
At the very least, they loaded all E one way and closed the magnetic circuit with all I stacked together, trusting natural punching irregularities to provide a built in gap.
next step up was to add a sheet of paper (newsprint is fine, better than nothing) to guarantee a minimum gap.
Just think about the "savings" in not doing so :O
Look at using mains toroidals if you want a cheap output transformer for push pull, excellent bandwidth, and great power handling. You need to oversize them a good bit, depending on desired bass response (50VA works for 20hz at 20-30~ or so watts max, depending on manufacturer) and watch the DC balance, but they are fantastic for this purpose.
Kodabmx, Shoog, and myself have played with them extensively and gotten very good results. I've played mostly with those made by Antek, kodabmx has used the triad units. High frequency performance extends way up above the audio band, and is quite linear.
EI core isn't very useful as outputs without basically stripping them and winding them new, been there, tried that.
Kodabmx, Shoog, and myself have played with them extensively and gotten very good results. I've played mostly with those made by Antek, kodabmx has used the triad units. High frequency performance extends way up above the audio band, and is quite linear.
EI core isn't very useful as outputs without basically stripping them and winding them new, been there, tried that.
The cheaper the 'power transformer', the likely 'better' the intrinsic gap for SE OPT use - such as using groups of E's interleaved with I's. Or as has been done with the downlight transformers, the B-H swing doesn't extend in to the saturation knee very far, and a '50W' 50Hz transformer is used for 10W of higher frequency signal transfer.
Nigel,
Check the prices. Almost can't beat that except if you use some pulled out transformer.
cheers
Transfer Multisort Elektronik - On-line Catalogue | 220 000 products offered.
Check the prices. Almost can't beat that except if you use some pulled out transformer.
cheers
Transfer Multisort Elektronik - On-line Catalogue | 220 000 products offered.
SE OPTs need a gap, as Juan Manuel pointed up.
PP OPTs from toroidal PT seems to me a not so good idea because of the huge area of the windings, this yields a big distributed capacitance, the core usually works near saturation, without a natural gap for any DC imbalance.
You can wind your own transformers, not so easy, but feasible, cheaper than commercial OPT and PT as well. Patience, perseverance and dedication, at the end you can do your own quality standards. The sky's the limit.
In the hope that this type of comment get's paraphrased - something like:
SE OPTs get more benefit from an introduced gap, or
SE OPTs achieve a better core utilisation when purposefully gapped, or
SE OPTs in general have a gap, or ....
Just saying, its not a black and white situation for a diy audio forum.
Hi,
I've also used toroid power transformers for PP OPT, I even have removed turns from the secondary to get exactly the impedance ratio. I haven't done that for my most ambitius projects but generally it works fine.
But there's another thing I would like to ask/test. Sometimes bulding a quick prototype costs close to nothing with inexpensive tubes and all those switch mode psus but OPT is a pain even if power transformers are used. So I was thinking if it's possible to get away with something like this: the pic is for a el34 with 2k OPT and 170V B+. But it could also be the same if the transformer was substibuted for a 2k power resistor and 360V B+. And signal could be taken from a small 1:1 - 8 ohm in, 8 ohm out - transformer placed in series with the resistor. I haven't tried this, sure a waste of power but do you think it has any meaning just for testing?
I've also used toroid power transformers for PP OPT, I even have removed turns from the secondary to get exactly the impedance ratio. I haven't done that for my most ambitius projects but generally it works fine.
But there's another thing I would like to ask/test. Sometimes bulding a quick prototype costs close to nothing with inexpensive tubes and all those switch mode psus but OPT is a pain even if power transformers are used. So I was thinking if it's possible to get away with something like this: the pic is for a el34 with 2k OPT and 170V B+. But it could also be the same if the transformer was substibuted for a 2k power resistor and 360V B+. And signal could be taken from a small 1:1 - 8 ohm in, 8 ohm out - transformer placed in series with the resistor. I haven't tried this, sure a waste of power but do you think it has any meaning just for testing?
Attachments
That would be me! I have 7 amps here all using Triad VPT series as OPTs. You need to balance the DC to avoid core saturation. I usually use CCS cathodes, but I've also used an automatic grid bias board from Pavel at audioamp.eu. You can also use "garter" bias.
Garter bias is my vote, and my usual solution. It works fantastically well, and is an olde schoole scheme, so it gives me that anachronistic warm and fuzzy feeling. I've stepped away from semiconductor active devices for many of my builds these days (other than for cathode follower loads, or power supply schemes) for (probably?) mostly psychological reasons, or whatever, so I like how simple it is.
No worse than most EI transformers for capacitance. You'll see in practice that the strays are a non issue at audio frequencies unless working at really high impedance, or when trying to use them as interstage coupling. Last time I ran frequency response measurements on several I found flat response out from 10Hz to 100khz, so I think that should be more than suitable
The only real gotcha with toroids is that they generally will not tolerate standing DC whatsoever, but that is easy enough to deal with.
I'm not much of an SE guy, so you're on your own if parafeed isn't your thing there.
The only problem with garter is the extra voltage wasted in the resistors, but sometimes B+ is too high and it's a good way to effectively lower it. The CCS bypassed with a large cap basically makes an automatic resistor, and for me, the sound is the same as if using a resistor instead.
For the record I've used standard fixed bias with the VPTs (each grid with it's own bias control) and it worked very well. I've even had a 20ma imbalance and the sound was still good.
For the record I've used standard fixed bias with the VPTs (each grid with it's own bias control) and it worked very well. I've even had a 20ma imbalance and the sound was still good.