Zero Feedback Impedance Amplifiers

15VA Toroid Input Transformer 1st Test

Hi Everyone,

I have done a quick test with the 15VA toroid transformer, and added the information to the toroid test amp page.

http://www.susan-parker.co.uk/zeus-toroid-amp-1.htm

To summarize:

For Vbias I used the centre tap on the 115 + 115 winding (which lowered the distortion slightly) rather than a center point on Rterm. So paying the extra for the double windings was worthwhile 🙂

Low level bandwidth is 10Hz to 25.5 kHz (-3 dB), 31.5 kHz (-6dB).

This is a little better then the 30VA toroid (which was 10Hz to 24 kHz (-3 dB), 28 kHz (-6dB) ) but I am still not seeing the 100's of kHz that other people have mentioned, even with these toroids being driven from a low impedance source.

Still the amplifier is more than adequate to listen to and enjoy 🙂

I hope to have some better testing facilities in a few days time with FFT plots and lots of other nice stuff.

Best wishes,
Susan.

http://www.susan-parker.co.uk/zeus.htm
 
Commentable thoughts

Hi SUSAN

Do u implemented INTERLEAVED WINDING SCHEME in ur Transformers to get maximum High Frequency Bandwidth. If Yes, Then How many Layers of Interleaved windings were used.

With Regards,
Workhorse Technologies.
 
Back to Susan -

I specifically wanted to use a smaller size of MOSFET than the output devices in the bias supply to cut down the amount of current in the bias chain - wasteful, and all that. I find that if you use a bias MOSFET one die size less than that of the output devices, you can adequately bias the output devices with 1/3 or less of the output device current flowing in the bias MOSFET. I have used this technique to advantage in MOSFET "diamond" buffer output stages for a couple of amplifiers I have built. In that application, you get thermally compensated bias and output MOSFET drive in one fell swoop, as long as you remember to mount the driver/bias MOSFETs on the same heat sinks as the outputs. In this case, I will be using an IRF531 to bias each set of IRF540 output devices. It will require about 30ma for an output bias of 100-200ma. I plan to use only a 30V supply, so 100V devices will have plenty of operating margin. My 44V toroids do appear to have bifilar output windings, so they should work OK as output transformers, as long as I can get good enough device matching to minimize any DC bias in the core. As an added bonus, The center tap on each toroid is already soldered together, so I don't have to worry about the push-pull phasing (That is, if I assume it's actually done right...).

I've been thinking about winding techniques for toroidal output transformers that might be easier to do for hand winding, yet incorporate most of the advantages of the multifilar technique. A fair compromise would be to use alternate layers of push-pull primary and single secondary windings, with the primary layers wound bifilar (or multifilar). One then connects the various layers together in series to get the final number of turns. This cuts down the amount of wire necessary to wind any given layer, which is an absolute boon if you are winding a toroid by hand. It also preserves the matching between the push-pull halves. This is what I will do if I get around to winding my output transformers (I have to find my toroids first - I just moved). To be honest, if the store-bought toroids sound good enough, I may not get that far... I have four other amplifiers sitting on my workbench in various stages of construction : two symmetric folded cascode designs with MOSFET diamond buffer output stages, a discrete "current feedback" design, and an open loop class A design inspired by the Pass Zen series. The push- pull amp gets temporarily bumped to the head of the line only because it's so simple, and because the transformers were right on hand. As you can guess, there will be a grand amplifier shoot-out when I finaly get all these beasts assembled. I have access to an HP4194A network analyzer and an Audio Precision Series 2 analyzer at work, so I will be able to chacacterize them once they are done. All these particular topologies simulated in Orcad as having nice output harmonic structures (low higher order distortion components, distortion dominated by 2nd harmonic). I want to find how close the actual amplifiers are to the simulations, and if there is any correlation to listenability. I have an open loop JFET preamp that measured out as being not too different from the simulation results. The next step is the always tedious one of putting it together with box and jacks and switches so I can audition it.
 
(... taking a moment or three from work to reply.)

Dear Circlotron,

Thank you for this information.

Circlotron said:

I am unable to view the images but from the description I would make the following response.

Irrespective of any other "prior art" I designed this amplifier from first principles. There is little new under the sun however and it is not surprising that there exists other versions prior to my own efforts.

I use a proper output transformer, with sectioned windings for load matching.

I do not see any mention of the importance of matching the number of turns between the two halves of the inductor (a generic mains voltage step-up/down autotransformer is suggested) yet the author is claiming an amplifier for low distortion.

I do not see any mention of multi-filar winding to improve the transformer action.

The patent says "The inductive reactance of the inductor provides a relatively high impedance to alternating current, and causes most of the alternating current output signal from the amplifier to be applied to the loudspeaker."

This is not a CORRECT description of the operation of the output stage, just some vague speak about current flow.

There is no mention of the inductor generating the effective second power rail. (Nor have I seen a clear description of this anywhere else for that matter.)

I see a specific mistake where it is claimed that a failure in a conventional amplifier can put DC across the loudspeaker but implies that with this version this does not happen. Sorry but this is a false claim. A failure in one of the devices can put DC across the loudspeaker.

Remember that a mosfet avalanches and goes short circuit if its maximum VDS is exceeded. This is not mentioned. Nor is the use of zeners to protect the gate from exceeding the VGS limit.

I use a separate secondary winding on my output which will prevent DC from being placed across the speaker under any condition.

The claim encompasses bipolar transistors but the base drive current required for this precludes the use of bipolar devices in this simple configuration for low distortion audio (FM RF power is a different matter).

It should also be noted in this context that there is the little matter of minority current in bipolar devices which also further reduces their suitability for this application.

I use an input transformer as a gain stage and to set the bias, not some scheme with resistors and capacitors.

A patent is meant to be method and application, with a clear description of the construction, operation or process of the invention so that someone competent in the general art can successfully understand and duplicate the object of the patent.

I do not see a clear description or a full understanding on the part of the author.

The current use of mains toroid transformer secondaries with a speaker across the two windings is a quick and convenient method of getting something working, but is not my preferred embodiment.

Overall I do not feel that this changes anything about my amplifier design.

Best wishes,
Susan.

P.S.
Replies to other posts this evening.
 
Picture in .gif format:

/Hugo
 

Attachments

  • us004096443-001.gif
    us004096443-001.gif
    74.8 KB · Views: 1,197
Susan-Parker said:
I am unable to view the images but from the description I would make the following response.

Alternatiff is an excellent, free, plugin for viewing ".tiff" images within any of the major web browsers (Internet Explorer, Mozilla, Opera, etc.). It automatically launches itself whenever certain filetypes are called.

It is especially small(memory wise), quick, and reliable. It also is much faster than Quicktime, which is usually launched by browsers.

I highly suggest this little jewel for anyone who spends a lot of time looking up patents at the US Patent Office.
 
Re: Commentable thoughts

Hi amp_man_1,

amp_man_1 said:
Hi SUSAN

Do u implemented INTERLEAVED WINDING SCHEME in ur Transformers to get maximum High Frequency Bandwidth. If Yes, Then How many Layers of Interleaved windings were used.

With Regards,
Workhorse Technologies.

Input:

The input transformer is interleaved and made in two identical halves, see pic on web pages...

http://www.susan-parker.co.uk/zeus-in-tx.htm

... which shows the dual section bobbin.

Effectively this makes the transformer

[ 1 : 5 ] + [ 1 : 5 ]

on the same core with each half would as

1/2 Secondary : Primary : 1/2 Secondary

therefor three layers.

The primaries can be wired in series, or in parallel for either an overall 1:5 (600 ohm) or 1:10 (150 ohm) setup up.

The reason for the extra complication it to have the secondary with a center point for the bias voltage. And therefor it is important that the winding of the two halves are matched as closely as possible both in number of turns and in resistance as the transformer is performing two functions.

On a toroid each half would be wound over a 180 degree arc. Note however that I have found with lower impedance driving that the Rterm is significantly reduced to level off the high end peak.

Output:

The output transformer is wound multi-filar as a single winding - with the secondaries split off as appropriate.

The output transformer is very specifically NOT wound in separate sections as would normally be usual in a valve output transformer.

Hope this helps.

Best wishes,
Susan.

P.S.

What value source resistors are you using to share the current out amongst your paralleled mosfets. I am guessing around 0.47 ohm?
 
Re: Re: Oh, I forgot....

Susan-Parker said:
Ta, now I can see this 🙂... and even better the references quoted.

Not bad, eh?

My browsers used to automatically try to use Apple's Quicktime to view patents and it was intolerably slow. It would completely bog down my entire computer. It took minutes to scroll down a single page. Very frustrating. :headbash:

Now, it's a cinch
 
Hi wrenchone,

Thank you for your long and informative post.

wrenchone said:
Back to Susan -

I specifically wanted to use a smaller size of MOSFET than the output devices in the bias supply to cut down the amount of current in the bias chain - wasteful, and all that. I find that if you use a bias MOSFET one die size less than that of the output devices, you can adequately bias the output devices with 1/3 or less of the output device current flowing in the bias MOSFET. I have used this technique to advantage in MOSFET "diamond" buffer output stages for a couple of amplifiers I have built. In that application, you get thermally compensated bias and output MOSFET drive in one fell swoop, as long as you remember to mount the driver/bias MOSFETs on the same heat sinks as the outputs. In this case, I will be using an IRF531 to bias each set of IRF540 output devices. It will require about 30ma for an output bias of 100-200ma.


This sounds like a sensible way to go. I went for the simple "if they are all off the same type, die, manufacturer, and date they should be close" approach 🙂

I will definitely try this as needlessly chucking heat out is not environmentally friendly.

I plan to use only a 30V supply, so 100V devices will have plenty of operating margin. My 44V toroids do appear to have bifilar output windings, so they should work OK as output transformers, as long as I can get good enough device matching to minimize any DC bias in the core. As an added bonus, The center tap on each toroid is already soldered together, so I don't have to worry about the push-pull phasing (That is, if I assume it's actually done right...).

I have run up to 100 mV differential without any apparent problems although it doesn't do anything for the distortion levels (I only had two of a particular - and expensive - mosfet to try out).

I try to match to under 10 mV difference. The closer the better.

I've been thinking about winding techniques for toroidal output transformers that might be easier to do for hand winding, yet incorporate most of the advantages of the multifilar technique. A fair compromise would be to use alternate layers of push-pull primary and single secondary windings, with the primary layers wound bifilar (or multifilar). One then connects the various layers together in series to get the final number of turns. This cuts down the amount of wire necessary to wind any given layer, which is an absolute boon if you are winding a toroid by hand. It also preserves the matching between the push-pull halves. This is what I will do if I get around to winding my output transformers (I have to find my toroids first - I just moved). To be honest, if the store-bought toroids sound good enough, I may not get that far...

A good approach, although what I would do is tri-filar wind say four equal sections. Two of the three wires per section are then the primaries, and the third becomes the secondary which can be series, series-parallel or parallel connected between the four sections.

This way all the windings are fully interleaved and each section is complete and it doesn't matter which of the three gets used for what.

This is for 1.0 mm wire and I therefor feel that tri-filar winding would be okay.

For my 35 watt transformer with 0.8 mm wire I was quad filar winding and then using doubled up on the secondary to get the DC and near DC resistance as low as possible to damp the loudspeaker bass to reduce ported "flapping" as at low frequencies the enclosure no longer provides resistance to the driver, unlike a fully enclosed box (which I personally prefer given the choice).

I have four other amplifiers sitting on my workbench in various stages of construction : two symmetric folded cascode designs with MOSFET diamond buffer output stages, a discrete "current feedback" design, and an open loop class A design inspired by the Pass Zen series. The push- pull amp gets temporarily bumped to the head of the line only because it's so simple, and because the transformers were right on hand.

Great 🙂

As you can guess, there will be a grand amplifier shoot-out when I finaly get all these beasts assembled. I have access to an HP4194A network analyzer and an Audio Precision Series 2 analyzer at work, so I will be able to chacacterize them once they are done. All these particular topologies simulated in Orcad as having nice output harmonic structures (low higher order distortion components, distortion dominated by 2nd harmonic). I want to find how close the actual amplifiers are to the simulations, and if there is any correlation to listenability. I have an open loop JFET preamp that measured out as being not too different from the simulation results. The next step is the always tedious one of putting it together with box and jacks and switches so I can audition it.

Any success in simulating my amp?

I look forward to your results. The chance to do comparisons will be very interesting.

Imagery. Can you test how far apart you can place the speakers before a hole opens up in the middle?

Amplifier signal to noise (max signal sine wave to unconnected input and residual noise.

Very encouraging.

Many thanks.

Best wishes,
Susan.
 
Re: Re: Re: Oh, I forgot....

Hi darkmoebius,

darkmoebius said:


Not bad, eh?

My browsers used to automatically try to use Apple's Quicktime to view patents and it was intolerably slow. It would completely bog down my entire computer. It took minutes to scroll down a single page. Very frustrating. :headbash:

Now, it's a cinch

Not bad at all.

I use Mozilla and it kept asking for a plug in and then not doing anything.

I don't have Quicktime on my portable as it interfears with the MACH2 CNC Gcode engine timing.

Best wishes,
Susan.
 
IRfan's website is gone!!!!!

Elso Kwak said:
Hi darkmoebius, I prefer Irfanview. It takes eveyrthing form TIF to GIF including JPEG.www.irfanview.com
:idea:

Luckily, you can still download it from Tucows

I will definetly check it out, although I prefer to use ACDSee Classic to view most other still picture file formats.

The advantage of Alternatiff is that it is an inline viewer, meaning that it is launched within the current browser's window(for those that don't know).

Although, the upside to IRfan is that handles almost all filetypes, so it is a "one size fits all" application - very handy.😀
 
Re: Re: Re: Re: Oh, I forgot....

Susan-Parker said:
Hi darkmoebius,



Not bad at all.

I use Mozilla and it kept asking for a plug in and then not doing anything.

I don't have Quicktime on my portable as it interfears with the MACH2 CNC Gcode engine timing.

Best wishes,
Susan.

Now I for one am confused. I am using Mozila Firefox and I did not realise that is was a picture anyway.
US patent office will only allow viewing the first page. Better use b2espacenet as I learned from King Gang Bang.
:clown:
Edit, that is:
http://l2.espacenet.com/dips/viewer?PN=US4096443&CY=gb&LG=en&DB=EPD
 
Re: IRfan's website is gone!!!!!

Elso Kwak said:
Actually the "bump" was referrring to the content of the document.😉 😉
I'm currently trying to read it. :cannotbe:

darkmoebius said:

The advantage of Alternatiff is that it is an inline viewer, meaning that it is launched within the current browser's window(for those that don't know).
Another advantage is that the pages can be printed, zoomed etc...

/Hugo 🙂
 
Re: Re: Re: Re: Re: Oh, I forgot....

Elso Kwak said:
Now I for one am confused. I am using Mozila Firefox and I did not realise that is was a picture anyway.

The US Patent office offers the entire patent form in a ".tiff" image form for viewing. If you look under the "Images" link, it will lead you there.

US patent office will only allow viewing the first page. Better use b2espacenet as I learned from King Gang Bang.
:clown:

I think the problem you are having is because IRfan is a viewer that is external to the browser, so you have to download each individual image in order to see them all. The menu on the blue bordered left side of the page provides links to view the "front page", "drawings", "specifications", and "claims". I assume that clicking these will launch the individual images in IRfan.

The big problem with that method is when viewing 10, 20, or 30 page documents. Very clinky.

On the other hand, if you use Alternatiff, it allows you to thumb through the entire document directly from the Patent webpage without any hassle by simply clicking the gold arrows on the left.

And as Netlist pointed out, it is very simple to zoom, print, save, etc. by simply "right-clicking" your mouse.

Download and install Alternatiff, I think you'll prefer it for this particular task. Be sure to set it to handle ".tiff filetypes during installation. Feel free to emial me if you need more help.

[And now, back to our regular programming]:wiz: