An audio amplifier I built doesn' work

Zip file has been attached.

Thanks!

I'm not sure about the FET models though. I think they model the crossover pretty well (the vertical MOSFET models do not) but I wouldn't bet on the capacitance.

Here's the measured open loop output impedance of Lateral & Vertical FETs plus a pair of BJTs (with some gm-doubling) thrown in. It's an upside down wingspan diagram, with gm instead of impedance.
 

Attachments

  • 2101 LATFET Zout 0R.png
    2101 LATFET Zout 0R.png
    87.7 KB · Views: 248
A buddy of mine uses ~30 of these amps to power his home theatre system. They work well.

I'm pretty sure his use a triplet of output devices, but that aside it looks like a very similar design.

I agree that taking this on as a first project is rather ... um ... "ambitious". :) I started with a simple Linn topology 15 W design.

Tom
 
Member
Joined 2018
Paid Member
Thanks!

I'm not sure about the FET models though. I think they model the crossover pretty well (the vertical MOSFET models do not) but I wouldn't bet on the capacitance.

Here's the measured open loop output impedance of Lateral & Vertical FETs plus a pair of BJTs (with some gm-doubling) thrown in. It's an upside down wingspan diagram, with gm instead of impedance.

My pleasure and thanks for the graphs. In the course of events I modeled amps with BJT and vertical FETs as well. When in doubt of a model, I set up a sim in such a way that the proper outcome should be near that of the producer's data sheet: for instance, capacity (in, out, reverse) can be checked by constructing a tuned circuit and observing frequency of resonance - this I can also verify with the spectrum analyzer.. Incorrect models also produce discontinuity in for instance THD vs V(in) curves. Notorious are depletion MOSFEts for CCS, that then have to be replaced by BJT equivalents. However at THD in the 0.01% and up THD range that won't be noticed.
 
Firstly let me say the Slone circuit does work, most of the time. I tried it and it worked fine but with a vas current of 38mA; Douglas Self tried it and it worked fine but with a vas current of a few hundred microamps; others have had various degrees of success plus a few total meltdowns along the way. The point here is that the circuit is not capable of replication with a defined vas current - sometimes you're lucky but more often you're not. The root of the problem is that the voltage at the junction of Q3,Q5 collectors and Q12 base is indeterminate ( original schematic parts numbering ). There are various ways of fixing this problem but they all involve impressing a known voltage at the junction mentioned above, directly or indirectly. The schematic shown below is a direct method which involves least change to the original schematic and pcb. The voltage impressed on the key junction must be greater than 1.2V, but not too much greater, to turn on the base/emitter junctions of Q12 and Q18. Replacing D1 and D2 by a red led provides a source of 1.7V and this is impressed on the key junction via R40 ( 10k). Adding R41 (51R) in the emitter of Q18 will result in a well defined vas current of about 10mA. This arrangement will cause a tiny imbalance in the Q5/Q6 current mirror but not enough to have any noticeable effect. Needless to say this arrangement must be mirrored in the upper half of the schematic. I changed the value of R21 to 2k as 100R seemed brutally low. I also changed the value of R10 to 200R to keep the ltp current source delivering 5mA. I have used this arrangement in real life amplifiers so I know that it works. On a more general point, discussion of this Slone amplifier, its topology, and related topologies has cropped up many times, over the years, on diyAudio. Maybe one of the moderators might care to look into the possibility of a sticky post on this topic, in order to ameliorate constant reinvention of the wheel?
 

Attachments

  • SloneMods.pdf
    9.8 KB · Views: 105
Last edited:
My pleasure and thanks for the graphs. In the course of events I modeled amps with BJT and vertical FETs as well. When in doubt of a model, I set up a sim in such a way that the proper outcome should be near that of the producer's data sheet.

Yes.

There are going to be a lot more graphs like that since I intend to measure the stuff no-one ever measures in an output stage, like what kind of current waveform it draws from the driving circuit, input capacitance vs current and how close to the rail voltage it is, bias drift, and stuff like that. I just got a multichannel soundcard for the occasion.
 
Only just seen this.
First reaction from post #1 - it should not work. Q13/Q18 complementary VAS have no means of limiting the current other than short circuit protection.
Bad design/
Recommend eliminating one half of the circuit as posted by peufue in #23 with one of the VAS transistors turned into a current source.
 
Member
Joined 2018
Paid Member
Yes.

There are going to be a lot more graphs like that since I intend to measure the stuff no-one ever measures in an output stage, like what kind of current waveform it draws from the driving circuit, input capacitance vs current and how close to the rail voltage it is, bias drift, and stuff like that. I just got a multichannel soundcard for the occasion.

Much appreciated as that also allows more verification of modeling. Even in Micro-cap it's possible to observe voltage dependent capacities. Exicon laterals have relatively low C (in), (out) and (rev) but the difference between N and P is about a factor 2 and when used in common source their influence is large.
 
Firstly let me say the Slone circuit does work, most of the time. I tried it and it worked fine but with a vas current of 38mA; Douglas Self tried it and it worked fine but with a vas current of a few hundred microamps; others have had various degrees of success plus a few total meltdowns along the way. The point here is that the circuit is not capable of replication with a defined vas current - sometimes you're lucky but more often you're not. The root of the problem is that the voltage at the junction of Q3,Q5 collectors and Q12 base is indeterminate ( original schematic parts numbering ). There are various ways of fixing this problem but they all involve impressing a known voltage at the junction mentioned above, directly or indirectly. The schematic shown below is a direct method which involves least change to the original schematic and pcb. The voltage impressed on the key junction must be greater than 1.2V, but not too much greater, to turn on the base/emitter junctions of Q12 and Q18. Replacing D1 and D2 by a red led provides a source of 1.7V and this is impressed on the key junction via R40 ( 10k). Adding R41 (51R) in the emitter of Q18 will result in a well defined vas current of about 10mA. This arrangement will cause a tiny imbalance in the Q5/Q6 current mirror but not enough to have any noticeable effect. Needless to say this arrangement must be mirrored in the upper half of the schematic. I changed the value of R21 to 2k as 100R seemed brutally low. I also changed the value of R10 to 200R to keep the ltp current source delivering 5mA. I have used this arrangement in real life amplifiers so I know that it works. On a more general point, discussion of this Slone amplifier, its topology, and related topologies has cropped up many times, over the years, on diyAudio. Maybe one of the moderators might care to look into the possibility of a sticky post on this topic, in order to ameliorate constant reinvention of the wheel?

One question, do I do it just on the bottom side or on the top side aswell?

I get a really loud high pitched sound from the speaker as soon as i turn on the power supply do you have any odea how i could eliminate that?
 
Hi,

What do you think about this schematics is it functional, has anyone built it, do you have any recommendations on what would you change. I pretty much gave up on the initial circuit utillizing mirror image input stage with current mirros, sience I tried most of the fixes people on this thread have recommended and it still doesnt work as it should. I just get a load beeping sound from the speakers.

Now I want to build a simmilar circuit so I can recycle components and the power supply. I am wondering if this circuit has any problems that I should look out for. Firstly I can see that this circuit still uses current mirrors I suppose that they must be stable with the use of matched transistors.

Other than that I could use some grounding tips and would like to know if the supply rails can be increased to +/- 60V.

I've also been looking into the optimos design as an option and it seems all fine but the diferential transistors MPS8099 and MPS8599 are unavailable in my country so are there any replacements, could I maybe use 2N5551 and 2N5401 or not?

Appreciate all the help and replies, thank youo very much. I really hope I can get one of these amplifiers to work.
 

Attachments

  • Zajeta slika.JPG
    Zajeta slika.JPG
    54.4 KB · Views: 143
i have but it and it works. However, i strongly suggest that you buy Rod Elliot's P101 boards from him at Elliot Sound Products. They are not expensive and the P101 works well.

He will provide you with constructor access to his website where you can get the real component values and constructor tips.

I have built 4 P101s and all operate from 56V rails with toroidal transformers rated from 300 to 1000 VA. You will need a soft start when using large toroidals to prevent the turn-on thump.
 
+1 for P101

The layout on your board looks exactly like the schematic, which is not a good idea for a power amp.

For example you want the power input wires near the output transistors, so the high current doesn't run through the whole board. Due to common impedance coupling and magnetic coupling, this can cause all sorts of problems like high distortion or even oscillation.
 
Alright, so I decided to do this design, sience i already have all the components needed.

I have a couple more questions before I get into building it:

Should I add catching diodes to the original schematics, what do you think of component values, are they reasonable or should they be changed to higher/lower values.

Do I connect the signal ground seperatly or together with the rest of the grounds. Do you maybe have any good examples of star grounding I dont quite understand how to implement it onto the PCB.

Thanks.
 

Attachments

  • Audio Amplifier 120W.pdf
    56.8 KB · Views: 74
That amp already has catching diodes across the rails to ground. It does not have any gate protection diode networks however. That Slone amp is not a good one for beginners. It also can be unstable and layout will affect that.

Why do you need such high power? If you have sensitive speakers then a 50 W amp that uses a single pair of Mosfets should generate enough sound. You would only need power rails of 40-45 VDC too.

Options:
1. The P101 will give you lots of power and is stable.

2. If you have most of the semis then you could instead build something like the amp based on the Philips design - see thread on this forum. You just get a LT1056 and some zeners and you will have all the semiconductors for that amp. I built one last weekend and it sounds very good. It runs cool, is small and with no signal is dead quiet.

3. Another option is the VSSA design which uses the same Lateral Mosfets and drivers. There are many threads on that design here with the PeeCeeBee being the simplest version of that concept.

No output inductor is needed for these amps, although I usually add one and use a speaker protection circuit to at least detect DC and shut off the speakers if DC is found.