Why not use a bigger laminationstack? That should lower distortion a lot, especialy at higher powerlevels! Same should work for the output-transformer. ???
Hi Circlomanen,
Thank you for your post.
In principle this is a good idea and would give the benefits as you suggest.
Unfortunately Mumetal is not available from Sowter in larger than 3/4" lamination sizes, and in this instance I was looking to see if there was any advantage to using Mumetal instead of the M6 silicon steel specifically in my input transformer.
Given that I am chasing 0.001% ths distortion levels the improvement of using a 50/50% mix would seem to be worthwhile for the relatively small additional cost in this size. Going for the 100% Mumetal is still economic but does have the drawbacks as I mentioned.
If one is intending to have a separate amplifier for lower bass (as I do) then a 100% Mumetal core isn't a problem.
An improvement could be obtained by having a deeper lamination stack in the same lamination size, but this would involve a redesign of the transformer. My current input transformer has a 3/4" stack, I believe bobbins are available for a 1-1/4" stack.
Unfortunately the high cost of Mumetal becomes prohibitive for larger lamination sizes (if they are available) that one would use for an output transformer. In this case I would be looking to move from M6 to M3 or specialty C cores.
Best wishes,
Susan.
Thank you for your post.
Circlomanen said:
In principle this is a good idea and would give the benefits as you suggest.
Unfortunately Mumetal is not available from Sowter in larger than 3/4" lamination sizes, and in this instance I was looking to see if there was any advantage to using Mumetal instead of the M6 silicon steel specifically in my input transformer.
Given that I am chasing 0.001% ths distortion levels the improvement of using a 50/50% mix would seem to be worthwhile for the relatively small additional cost in this size. Going for the 100% Mumetal is still economic but does have the drawbacks as I mentioned.
If one is intending to have a separate amplifier for lower bass (as I do) then a 100% Mumetal core isn't a problem.
An improvement could be obtained by having a deeper lamination stack in the same lamination size, but this would involve a redesign of the transformer. My current input transformer has a 3/4" stack, I believe bobbins are available for a 1-1/4" stack.
Unfortunately the high cost of Mumetal becomes prohibitive for larger lamination sizes (if they are available) that one would use for an output transformer. In this case I would be looking to move from M6 to M3 or specialty C cores.
Best wishes,
Susan.
Susan,
You may have already mentioned this and I missed it: Where do you source your mumetal laminations, and what's the approximate cost?
You may have already mentioned this and I missed it: Where do you source your mumetal laminations, and what's the approximate cost?
Mumetal laminations
Hi Bill,
I bought my 3/4" Mumetal laminations from Sowter as I already have transformers from them (# 8160).
http://www.susan-parker.co.uk/zeus-in-tx.htm
It costs about 40 to 50 USD per inch, depending on if one has to pay sales tax.
Best wishes,
Susan.
Hi Bill,
Bill F. said:
I bought my 3/4" Mumetal laminations from Sowter as I already have transformers from them (# 8160).
http://www.susan-parker.co.uk/zeus-in-tx.htm
It costs about 40 to 50 USD per inch, depending on if one has to pay sales tax.
Best wishes,
Susan.
Can I use a larger M-6 laminationstack and get better results than a smaller mu-metal stack??
If I use a EI-180 stack in an 40 watt version (as a output-transformer) , wouldnt that have lower distorsion than an EI-96??
What happens if I use a 10 inch (25 cm) thick stack of EI-180 laminations??
Hi Circlomanen,
The problem is that M6 has higher losses than Mumetal and so there is an intrinsic point where performance can't increase for a given driving impedance.
As I am driving my output transformers from a low impedance source I am operating under nearly ideal conditions.
What is interesting is to look at the possibility of going for a hugely oversized core and then using an air gap to end up with a similar inductance to the current output transformer. Depending on what one reads this will make performance better or worse!
I have some M6 EI 2-1/8" laminations for this purpose and 3" bobbins to match. Wire for this is going to be over UKP 50. And the transformer will be heavy!
Just adding core size on it's own doesn't necessarily improve matters as the inductance goes up (which affects the HF response). So one uses thicker wire with less turns to compensate. Inductance is reduced but the DC resistance also goes down. So the mosfets become more twitchy to bias into their "nice" zone.
Adding the gap reduces the inductance so allows one to use only slightly heavier gauge wire with more turns. I aim to have a DC resistance of about 0.5 ohms or a little higher for each primary arm.
Whether this transformer will actually result in lower distortion will be interesting to see. I suspect for PP it won't be as good but for SE it will probably be better.
Best wishes,
Susan.
Circlomanen said:
The problem is that M6 has higher losses than Mumetal and so there is an intrinsic point where performance can't increase for a given driving impedance.
As I am driving my output transformers from a low impedance source I am operating under nearly ideal conditions.
What is interesting is to look at the possibility of going for a hugely oversized core and then using an air gap to end up with a similar inductance to the current output transformer. Depending on what one reads this will make performance better or worse!
I have some M6 EI 2-1/8" laminations for this purpose and 3" bobbins to match. Wire for this is going to be over UKP 50. And the transformer will be heavy!
Just adding core size on it's own doesn't necessarily improve matters as the inductance goes up (which affects the HF response). So one uses thicker wire with less turns to compensate. Inductance is reduced but the DC resistance also goes down. So the mosfets become more twitchy to bias into their "nice" zone.
Adding the gap reduces the inductance so allows one to use only slightly heavier gauge wire with more turns. I aim to have a DC resistance of about 0.5 ohms or a little higher for each primary arm.
Whether this transformer will actually result in lower distortion will be interesting to see. I suspect for PP it won't be as good but for SE it will probably be better.
Best wishes,
Susan.
I just got my hands on a big circuit board containing 48 HUF76143P3 mosfets, I think its old military stuff from a huge power inverter to get ac from 24 volts.
Here is the datasheet, although I don't understand everything it says it seems like it has very high gain, although I don't know if the same applies at these low currents, I don't know if its linearity is good or bad either.
So I just build it with a 14 cm² core transformer having 70 turns for each winding, it sound surprisingly good.
Now all I need is a good gain stage because I don't want to use a transformer at the input.
Idle current is set at either 100 mA or 500 mA, my multimeter shows different values at different settings lol.
Here is the datasheet, although I don't understand everything it says it seems like it has very high gain, although I don't know if the same applies at these low currents, I don't know if its linearity is good or bad either.
So I just build it with a 14 cm² core transformer having 70 turns for each winding, it sound surprisingly good.
Now all I need is a good gain stage because I don't want to use a transformer at the input.
Idle current is set at either 100 mA or 500 mA, my multimeter shows different values at different settings lol.
Hi Bobo1on1,
Thanks for your post 🙂
I have found relatively small difference between mosfet types. Its worth matching pairs if one can. And some are a bit better than others but there does not seem to be a huge difference except for the top end when the part is approaching the power rails.
The comparison graphs I did for the SE version show this.
I assume you are using quite a low voltage power supply as these parts have a 30 volt VDSS?
Thanks for your post 🙂
bobo1on1 said:
I have found relatively small difference between mosfet types. Its worth matching pairs if one can. And some are a bit better than others but there does not seem to be a huge difference except for the top end when the part is approaching the power rails.
The comparison graphs I did for the SE version show this.
I assume you are using quite a low voltage power supply as these parts have a 30 volt VDSS?
Here's a webcam pic of how its sitting on my desk right now, not much to see because of all the wires.
The speaker is connected directly to the primaries, I wound the transformer with 4 strand telephone wire but when I use the secondairies the highs don't sound as nice.
There is also a ptc in series with the 5k pot for thermal stability, the ptc is about 60 ohms at room temperature and about 700 ohms when the heatsink gets really hot, idle current only goes up a few miliamps.
It's all running off a 16 volts rail, including the pot for bias.
The speaker is connected directly to the primaries, I wound the transformer with 4 strand telephone wire but when I use the secondairies the highs don't sound as nice.
There is also a ptc in series with the 5k pot for thermal stability, the ptc is about 60 ohms at room temperature and about 700 ohms when the heatsink gets really hot, idle current only goes up a few miliamps.
It's all running off a 16 volts rail, including the pot for bias.
I am still working on breadboarding one of these amplifiers using torioidal 60Hz transformers and IRF540 MOSFETs. I may try use a discrete unbal to balanced converter/driver intead of the input transformer.A question to Susan.... What is your setup for using your sound card and software to measure distortion? I have WinAudio MLS Pro 1.8, and want to put it to work. I was using the Audio Precision analyzer at work, but it's, shall we say, under the weather....
Hi bobo1on1 ,
Many thanks for the pic ...
... and further details.
Looks good.
Best wishes,
Susan.
Many thanks for the pic ...
bobo1on1 said:
... and further details.
Looks good.
Best wishes,
Susan.
Hi Wrenchone,
Glad to hear from you.
Oops!
My setup is an E-Mu 1212M soundcard with stereo balanced line in and out.
I mostly use low out (-10 dBV), high in (+4 dBu) settings.
For single stage Zeus amplifier I have a line driver using the TI TPA6120 headphone amp (PA's QRV07 is very useful here). I am only using one side per channel as it doesn't like driving balanced across both sides i.e. the TPA6120 outputs need a ground reference to drive into.
For two stage Zeus I connect the sound card output directly to the input transformer (e.g. the Gardners 1:4 600R:10K).
For measurement I connect the soundcard input directly to the load as I don't want to put in any further stages of possible distortion.
I have bought a new set of non-inductive power resistors to make up a new load. It will have four series 1.00 ohms at the bottom so I can tap off a divided voltage at higher powers to keep within the max input on the soudcard. I have various values so I can increment the load up to 60 ohms to plot distortion versus load (in all I have 5 x 1R plus 5R, 10R, 15R, 22R, 33R, 47R and 56R).
Unfortunately I am not completely happy with the E-Mu 1212M as it has a fair amount of residual distortion (at -120 dB) and I can only use one channel input as there is a lot of crosstalk between channels.
This is an issue for me as I have a noise floor at -150 dB.
This is why I include loopback measurements in my pages to show which of the stuff is from the measurement system, not the amp.
I also have problems with the AISO drivers and if I am not very careful the program crashes and the computer then has to have a reboot to clear the AISO driver. If I always stop the program from acquiring before changing any settings I am usually okay.
The E-Mu 1212M was the best 24bit 192kHz PCI sound card with balanced I/O that I could find/afford. But now I am thinking that I need to use a separate external analog I/O module and customize it specifically for measurements.
Maybe I should be looking for some external Firewire unit? That or getting the external E-Mu box and hacking it.
Hope this helps 🙂
Best wishes,
Susan.
P.S. I just got the new 1.90 version of WinAudio MLS this morning, haven't had a chance to look at it yet.
Glad to hear from you.
wrenchone said:
Oops!
My setup is an E-Mu 1212M soundcard with stereo balanced line in and out.
I mostly use low out (-10 dBV), high in (+4 dBu) settings.
For single stage Zeus amplifier I have a line driver using the TI TPA6120 headphone amp (PA's QRV07 is very useful here). I am only using one side per channel as it doesn't like driving balanced across both sides i.e. the TPA6120 outputs need a ground reference to drive into.
For two stage Zeus I connect the sound card output directly to the input transformer (e.g. the Gardners 1:4 600R:10K).
For measurement I connect the soundcard input directly to the load as I don't want to put in any further stages of possible distortion.
I have bought a new set of non-inductive power resistors to make up a new load. It will have four series 1.00 ohms at the bottom so I can tap off a divided voltage at higher powers to keep within the max input on the soudcard. I have various values so I can increment the load up to 60 ohms to plot distortion versus load (in all I have 5 x 1R plus 5R, 10R, 15R, 22R, 33R, 47R and 56R).
Unfortunately I am not completely happy with the E-Mu 1212M as it has a fair amount of residual distortion (at -120 dB) and I can only use one channel input as there is a lot of crosstalk between channels.
This is an issue for me as I have a noise floor at -150 dB.
This is why I include loopback measurements in my pages to show which of the stuff is from the measurement system, not the amp.
I also have problems with the AISO drivers and if I am not very careful the program crashes and the computer then has to have a reboot to clear the AISO driver. If I always stop the program from acquiring before changing any settings I am usually okay.
The E-Mu 1212M was the best 24bit 192kHz PCI sound card with balanced I/O that I could find/afford. But now I am thinking that I need to use a separate external analog I/O module and customize it specifically for measurements.
Maybe I should be looking for some external Firewire unit? That or getting the external E-Mu box and hacking it.
Hope this helps 🙂
Best wishes,
Susan.
P.S. I just got the new 1.90 version of WinAudio MLS this morning, haven't had a chance to look at it yet.
Hi, I'm butting in late here - having spent the last day reading the previous postings. 
So, any update with the twinax transformer? It seemed to go very quiet after a rather enthusiastic first report.
Sound IO. Funny I have been thinking about the same issue. However I am a Mac and Unix guy. So anything from Creative is a waste of time. However Avid are nice people, and these things look very interesting.
http://www.m-audio.com/products/en_us/FirewireAudiophile-main.html
http://www.m-audio.com/products/en_us/AudiophileUSB-main.html
But this guy looks almost spot on if you want balanced out.
http://www.m-audio.com/products/en_us/FireWireSolo-focus.html
I'm very interested in pursuing more esoteric distortion metrics, so rolling one's own code, and using a good sound IO device seem the only way.
So, any update with the twinax transformer? It seemed to go very quiet after a rather enthusiastic first report.
Sound IO. Funny I have been thinking about the same issue. However I am a Mac and Unix guy. So anything from Creative is a waste of time. However Avid are nice people, and these things look very interesting.
http://www.m-audio.com/products/en_us/FirewireAudiophile-main.html
http://www.m-audio.com/products/en_us/AudiophileUSB-main.html
But this guy looks almost spot on if you want balanced out.
http://www.m-audio.com/products/en_us/FireWireSolo-focus.html
I'm very interested in pursuing more esoteric distortion metrics, so rolling one's own code, and using a good sound IO device seem the only way.
OK, first, greate thread Susan. Read the whole thing over the last two days. Looked through the stash, found 6 IRF450s an old, dead, Phase Linear 400 (for those not familiar, a '70s 200WPC bipolar power amp), some 15VA toroids & some 225VA toroids, and got to work. Fortunately, the Phase Linear power transformer actually had two 60V secondaries instead of one center-tapped one, so I was able to parallel them and end up with a 78V B+. A bit higher than I woulda liked, but the IRF450s can handle 500V... Built the whole thing with parts I had on hand (which is why it's currently zenerless, didn't have any of those, BTW, how important is the power rating on the zener's?). I was able to match up two pairs of the IRF450s (but the pairs weren't even close to each other, so I had to implement seperate bias adjustments for each channel) which was fine until I toasted a pair with a careless meter probe 
. So, now it's time for some more FETs. Anyone have any ideas for a 200V+ TO3 FET? I can probably find some more IRF450s, but if there are better choices, I might was well make the change now.
Did have it making music before dinner, but havn't had a chance to listen seriously. Need to find my decent portable CD player, was using the headphone output on what has to be the cheapest portable CD player ever before dinner, still sounded fairly decent, more on the sound to follow tomorrow... Maybe this is the excuse I need to buy an iPod 😉
Peace
. So, now it's time for some more FETs. Anyone have any ideas for a 200V+ TO3 FET? I can probably find some more IRF450s, but if there are better choices, I might was well make the change now.Did have it making music before dinner, but havn't had a chance to listen seriously. Need to find my decent portable CD player, was using the headphone output on what has to be the cheapest portable CD player ever before dinner, still sounded fairly decent, more on the sound to follow tomorrow... Maybe this is the excuse I need to buy an iPod 😉
Peace
Back to Susan - I'm currently using a Terratec DMX 6Fire in my main system, which is a good quality consumer card with a set of drive bay I/Os. Unfortunately, everything's unbalanced, but I'd have to go to a much more expensive pro-level card to get balanced I/O. In my music system (currently apart), I have one of the old Event Gina cards, which I found to be exceptionally quiet, at least from a subjective standpoint. I never tried measuring the absolute noise floor, though. The setup was pretty stable running with Cool Edit Pro or Soundforge 4. Unfortunately, a computer is a really stinky environment for running sensitive analog gear... Maybe a firewire or USB external solution would be quieter.
I assume you bought the 24 bit signal generator plugin to drive the sound card outputs. I currently have the free dowload generator, but will probably end up springing for the better quality 24 bit software.
I assume you bought the 24 bit signal generator plugin to drive the sound card outputs. I currently have the free dowload generator, but will probably end up springing for the better quality 24 bit software.
Francis_Vaughan said:
Chris Liscio ( FuzzMeasure) has at least one m-audio devices... except for the flakiness of the drivers he has made some favorable comments recently (M-Audio FireWire 1814).
http://www.supermegaultragroovy.com/
dave
Roscoe Primrose said:
Roscoe,
Do you really need an excuse. Where else can you get a portable back-up device capable of storing names/addresses (amounst many other capabilities) that comes with Wolfsen DACs?
dave
lohk said:
It is amazing how selling millions of units pushes the price down -- o all our benefit.
dave