Zero Feedback Impedance Amplifiers

[mikeks]

Originally posted by Susan-Parker
Test at 1 kHz.

No filter (test bandwidth is 500 kHz).

With no filter, performance is excellent for such a simple design....even at 1KHz.....

Assuming of course, that there are no nasty surprises at HF with non-standard loads....

[Graham Maynard]

Hi Susan,

The mid range distortion figures show that the amplifier is 'characterising' the sound. This is where fundamental nulling will reveal what the amplifier is adding.

Cheers ........... Graham.

[Susan-Parker]

MOSFETs IRFP150
Supply 34.00
Vbias 3.88
Q1 bias 0.217
Q2 bias 0.233
Load 8.02
Frequency 1.00 kHz

Watts AC-RMS DIST%
40.4 18.00 0.340
36.1 17.03 0.268
32.3 16.11 0.250
28.5 15.11 0.305
24.6 14.06 0.374
21.2 13.04 0.411
18.0 12.01 0.449
15.4 11.13 0.485
12.5 10.00 0.526
10.1 9.00 0.570
7.98 8.00 0.616
6.18 7.04 0.654
4.56 6.05 0.679
3.20 5.07 0.684
2.00 4.01 0.612
1.18 3.08 0.437
0.78 2.50 0.275
0.50 2.00 0.156
0.38 1.75 0.115
0.28 1.50 0.076
0.19 1.25 0.048
0.12 1.00 0.034
0.10 0.90 0.030
0.080 0.80 0.026
0.061 0.70 0.027
0.045 0.60 0.016
0.031 0.50 0.028
0.020 0.40 0.041
0.011 0.30 0.028
0.008 0.25 0.030
0.005 0.20 0.030
0.003 0.15 0.055
0.001 0.10 0.115

Notes:

This test uses the original ST brand IRFP150 mosfets as fitted to the amp in the picture.

The mid power bump is at a little higher power than that with the IR IRFP140, but slightly lower in distortion.

The amplifier starts becoming power supply limited at 17 volts RMS.

100 Hz distortion figures are similar or slightly lower.

Tested with 500 kHz measurment bandwidth.

========

Low frequency power curve - Hz versum maximum AC-RMS at that frequency.

Load 8.02

Hz AC-RMS Watts
10.0 8.03 8.036
11.0 8.69 9.411
12.0 9.20 10.548
13.0 9.90 12.215
14.0 10.51 13.766
15.0 11.03 15.162
16.0 11.76 17.235
17.0 12.29 18.824
18.0 13.10 21.387
19.0 13.60 23.051
20.0 14.10 24.777
21.0 14.64 26.711
22.0 14.93 27.780
23.0 15.40 29.556
24.0 16.14 32.465
25.0 16.39 33.479
26.0 17.05 36.229
27.0 17.55 38.385

The above is limited by the line driver's ability to drive the input transformer.


=====

4 ohm load distortion figures are higher (this transformer does not have split secondaries - so it is not properly matched to this lower impedance).

Load 4.00
Frequency 1.00 kHz

Watt AC-RMS DIST%
25.0 10.00 1.853
20.3 9.00 1.302
16.0 8.00 0.768
12.3 7.00 0.846
9.0 6.00 0.933
6.3 5.00 1.031
4.0 4.00 1.135
2.3 3.00 1.232
1.6 2.50 1.230
1.0 2.00 1.094
0.77 1.75 0.942
0.56 1.50 0.727
0.39 1.25 0.408
0.25 1.00 0.282
0.20 0.90 0.194
0.16 0.80 0.152
0.12 0.70 0.122
0.09 0.60 0.089
0.06 0.50 0.072
0.040 0.40 0.060
0.023 0.30 0.044
0.016 0.25 0.027
0.010 0.20 0.011
0.006 0.15 0.010
0.003 0.10 0.058

=====

Best wishes,
Susan.

[Susan-Parker]

Hi,

Originally posted by mikeks


With no filter, performance is excellent for such a simple design....even at 1KHz.....

Assuming of course, that there are no nasty surprises at HF with non-standard loads....

Good

The figures do not change when I use the 80 kHz or 30 kHz low pass filters.

Once one has the input transformer any HF is filtered out.

... but it is important to know how the circuit sections behave.

I was involved in EMC/RFI conducted and radiated emissions testing of a battery/mains powered smoke alarm recently. All very low power stuff and a low few hundred hertz tick for its internal function.

Perfectly safe one would assume, and no emissions to worry about.

However it nearly failed the radiated emissions tests. Whilst in quiescent mode there wasn't a problem, but when the smoke unit was alarmed i.e. sounding it's pizzo horn, it was radiating up into several megahertz at a figure than was 80% of the test level failure point!

So much for low power and innocuous!

Best wishes,
Susan.

[bear]

Susan,

I've followed this thread with some interest...

Of course, you've fed bits and pieces out to the hungry little birdies!

It would be nice if ur webpages had an index so that someone could look at the parts of the design? If there is one, I don't know where to find it... certainly not from your home page?

With regard to the transformer specs, imho it's a bit like cheating to lightly load the transformer and test it. There's a certain "toroidal" output transformer company that shows rather remarkably good specs, but if you read the fine print they're testing it at a fraction of the practical operational levels for core saturation.

In general, for everyone else reading along, the multi-filar wound output transformer in the *low impedance* & *low turns ratio* variety tend to have substantially better resulting specs than do the usual output transformers. Typically bandwidths out to 250kHz. are possible whereas doing the same thing with even relatively low plate Z output transformers (like for a 300B) is not quite so easy, if acheivable at all.

No doubt this design has all the hallmarks of the most popular "DIY" style amps - like the original ZEN. What are those? Almost no parts, nothing to align or test, and ultimate simplicity. Hey, that's good!

The rub comes in the details - while almost anyone can build such amps, and they do make sound (sometimes good sound), there are usually several points of detail that do require some careful control and measurement to make the final practical project operate as well as the expectation of the "test bench" results indicate. This one is no different in that regard.

Among the points that need to be addressed still are:
- the effect of DC offset (device to device)
- the effect of bias offset to correct the DC offset wrt linearity
- what the optimal bias point is on the curve
- effect on power and frequency response of non-flat impedance loads
- and some other stuff that escapes me at the moment

Wrt the square wave response - I would not be satisfied with that result in any amp that came across my bench. It holds potential, but that ringing (imho) is a defict that is audible. Damping it might screw the overall response, or it might not depending on how or what the cause is/was... is it inherent to the output xfmr or? Important detail.

The use of unity coupled output transformers is nothing new, in case anyone thinks otherwise.

What's still not clear to me from a casual following of the thread is the design bias points, and how this design's power output varies with load (with and without the restrapped secondary) -
or is it truly a "current amplifier" at all?

But there is food for thought here...

Put up an index page, please?

Regards,

_-_-bear

http://www.bearlabs.com

PS. I have a precision coil winding machine here (USA)

[Susan-Parker]

Hi,

Originally posted by Graham Maynard
Hi Susan,

The mid range distortion figures show that the amplifier is 'characterizing' the sound. This is where fundamental nulling will reveal what the amplifier is adding.

Cheers ........... Graham.

Yes, it is interesting how the bump happens and then falls back to lower levels until the power rail starts to become a limiting factor.

From valve triode data I would expect it to just keep on rising - so I guess there are two data sheet graphs interacting here.

There is a small but noticeable difference between the IRFP140 and IRFP150 device characteristics. This indicates that it should be possible to optimize device type, load and power to maximize the performance of the amplifier to a given speaker.

Perhaps not so practical in a commercial product but less of a problem if one is building for oneself.

Thanks.

Best wishes,
Susan.

P.S. I would note that the IRFP150s are less well matched than the IRFP140s.

[Susan-Parker]

Hi Bear,

Originally posted by bear
Susan,

I've followed this thread with some interest...

Of course, you've fed bits and pieces out to the hungry little birdies!

Thank you for your interest and post.

I too have sought "clarity".

I would mention that I have not been deliberatly feeding out bits n' peices. I worked on the audio bits up to a decade ago and whilst many people have good memory for things I don't. I have also moved several times, ditiching at least 90% of my reference material, and it is a wonder to me that I have anything at all remaining from that time.

People have asked questions and I have been going away and actually doing measuments, to the best of my ability and with what I have to hand, in order to give an answer.

I have added the other pages to give background and scope to why I got started on my amplifier design in the first place.

It would be nice if ur webpages had an index so that someone could look at the parts of the design? If there is one, I don't know where to find it... certainly not from your home page?

There is a links section in the zeus page near the bottom to the other pages.

I haven't linked this stuff from my homepage as the amplifier discussion was in this forum, and I have been creating the additional web pages on the fly during the course of this last week.

Now that I have done this, when I have time, I may well put up a past projects page along with some other things I have done.

With regard to the transformer specs, imho it's a bit like cheating to lightly load the transformer and test it. There's a certain "toroidal" output transformer company that shows rather remarkably good specs, but if you read the fine print they're testing it at a fraction of the practical operational levels for core saturation.

It is not cheating to do some tests if the conditions of the tests are specified - there was not any small print. Also I would query the case of the load which is 50 ohms and at these higher frequencies it is not unlikely that the speakers could also be in that ballpark.

It may be less relevant to audio but it is relevant to compliance testing where one has to demonstrate from 100kHz (or lower) to 2 GHz (or higher) the conducted and radiated conditions for the amplifier.

In Europe there are stringent rules for the assembly and use of electronics commercially and although most people are building for personal hobby use it is prudent to make sure that one isn't transmitting somewhere in the VHF or UHF bands where speaker cables and the like are more than adequate as antenna.

In general, for everyone else reading along, the multi-filar wound output transformer in the *low impedance* & *low turns ratio* variety tend to have substantially better resulting specs than do the usual output transformers. Typically bandwidths out to 250kHz. are possible whereas doing the same thing with even relatively low plate Z output transformers (like for a 300B) is not quite so easy, if acheivable at all.

Yes, but there is little reference to this in the literature (that I knew of). I have clearly explained the importance of the multi-filar windings to the sucess of building this amplifier.

No doubt this design has all the hallmarks of the most popular "DIY" style amps - like the original ZEN. What are those? Almost no parts, nothing to align or test, and ultimate simplicity. Hey, that's good!

Thank you - however I hear a BUT in there somewhare...

The rub comes in the details - while almost anyone can build such amps, and they do make sound (sometimes good sound), there are usually several points of detail that do require some careful control and measurement to make the final practical project operate as well as the expectation of the "test bench" results indicate. This one is no different in that regard.

I am using standard off the shelf components with the exception of the input transformer, which I would and have admitted is a little more difficult to make, most is readily to hand. The output transformer bits are perhaps not quite so easy to get but I am sure this is not a problem to the valve purists to advise us on.

My test bench results are vanilla and anyone building my design with the components and levels described should get equal or better results.

My line driver to the input transformer is now obviously the limiting factor, and this needs to be addressed.

However the amplifier works quite happily from the headphone output of a CD player or walkman/iPod. So it is easy to get a basic rig running and then be able to spend time working on the preamp side of things in the knowledge that the overall system will be working at the end of the day.

Whilst on paper the quite high distortion levels might seem off-putting they need to be equated with tride valve amplifiers, not semiconductor ones.

Among the points that need to be addressed still are:
- the effect of DC offset (device to device)
- the effect of bias offset to correct the DC offset wrt linearity
- what the optimal bias point is on the curve
- effect on power and frequency response of non-flat impedance loads
- and some other stuff that escapes me at the moment

Yes, and this is all part and parcel of somehting called "DIY Audio".

But to give you some answers:

1. One needs to match the device as well as one can. I would suggest to aim for better than 10 mV matching at a bias of 225 mV at each primary of the transformer with respect to ground.

2. See 1 above. Seperating out the bias levels and adjusting them seperatly doesn't seem to make improvments.

3. Optimum bias point depends on a number of things including the device type - the bais levels I am using are at anywhere from 3.5 to 5.1 volts depending on the mosfet. Higher levels van be used with bigger heatsinks, a matter of experimantation. Big heatsinks are very expensive to buy, so a lower bias and therefor lower heat disapation is an advantage.

4. Non flat impedance loads will have effects on any amplifier. Some speakers will work better than others, just the same as with any system. I would look at types favioured by the low power single ended triode brigade as a starting point.

BTW The amplifier is designed for normal listning in personal environments, not for Carnige Hall, so I am not trying to address higher power levels.

Wrt the square wave response - I would not be satisfied with that result in any amp that came across my bench. It holds potential, but that ringing (imho) is a defict that is audible. Damping it might screw the overall response, or it might not depending on how or what the cause is/was... is it inherent to the output xfmr or? Important detail.

I agree, I am not happy either. However at least half the problem is in the line driver to the input transformer and as such until I can sort out something better for that there is no point in fiddling with the output side of things.

The use of unity coupled output transformers is nothing new, in case anyone thinks otherwise.

I have never claimed this.

What's still not clear to me from a casual following of the thread is the design bias points, and how this design's power output varies with load (with and without the restrapped secondary) - or is it truly a "current amplifier" at all?

The bias point is where I found that it works, without overly heating up.

I don't have the equipment to do the proper measurments to be able to answer the power output bit.

It may not be a "current amplifier" but this is how I understand transformers to operate. Maybe I am wrong in this detail, but it is better than describing the amplifier as a "flux convertor".

I come from a digital and instrumentation/control electronics background - so my understanding of these definitions may be different.

But there is food for thought here...

Put up an index page, please?

Regards,

_-_-bear

http://www.bearlabs.com

PS. I have a precision coil winding machine here (USA)

Ah, is that an offer to supply transformer bits - or kits for that matter?

As a matter of curiosity what are the performance figures for your single ended mosfet amp?

Many thanks.

Best wishes,
Susan.

[bear]

Damn!

I wrote a long reply and the forum software dumped it!!

Grrrrrrrrr.....

GRRR GRRR GRRRR

_-_-bear

[Susan-Parker]

Hi Bear.

Originally posted by bear
Damn!

I wrote a long reply and the forum software dumped it!!

Grrrrrrrrr.....

GRRR GRRR GRRRR

_-_-bear

Sory to hear that, these things can be quite frustrating.

Perhaps a synopsis of your comments?

Best wishes,
Susan.

[Graham Maynard]

I've lost posts due to software too.

I now write the entire message in MS wordpad, then copy/paste, and I do not delete my copy until I have seen the entry at diyAudio.

Cheers ............. Graham.