First solid state amp design/build, advise

[nplez1]

Better but still -
The output MOSFETs are drawn as bipolars.
The lower VAS connection is drawn over R32's GND.
As to the actual circuit.
The input LTP current is lower than recommended. Read D. Self for this.
The MOSFET stoppers could probably be lower. Cordell is the reference for MOSFETs.
The compensation capacitors could be swapped. Harry Dymond here.

Best wishes
David

I have one of Cordell's books, although I found the material to be a bit dense at times since I don't have a super extensive background in electrical engineering and theory. I'll have to look into Self and Dymond.

Thanks for the pointers!

 

[Dave Zan]

None of this needed

It is not needed, no.
But the OP asked for comments so I think he wants to learn and not just copy, and these are improvements with no real downside.
An increase in LTP current will improve the linearity and decrease the noise at no extra cost or complexity. Lower noise and distortion sounds better to me.
The swap of the compensation capacitors will improve linearity and increase slew rate also at no extra cost or complexity. Why use the inferior circuit just because someone (Slone?) made an ill informed decision? Since then Harry Dymond has published an analysis that explains that the less obvious choice is better.

Best wishes
David

Layout. Important that the stopper resistors should be as close as possible to the FETs to minimize stray inductance.
Cordell once more.

 

[homemodder]

Doing these mods will need redesign of compensation and hopefully the poster has the equipment to test stability once these mods are made. With this design stability is easier compromised because of the unusual cascode used.
Slone didnt make ill informed decisions at all, try smaller gate resistors as suggested by you on the original and you will be rewarded with oscilation in MHZ range. He was on top of his game, these new compensation methods are discoveries made a short time ago. Why not TMC, it yields even better results ??

 

[Dave Zan]

Doing these mods will need redesign of compensation and hopefully the poster has the equipment to test stability once these mods are made. With this design stability is easier compromised because of the unusual cascode used.
Slone didnt make ill informed decisions at all, try smaller gate resistors as suggested by you on the original and you will be rewarded with oscilation in MHZ range. He was on top of his game, these new compensation methods are discoveries made a short time ago. Why not TMC, it yields even better results ??

The poster has decided to do his own board layout and therefore will need to test stability anyway so I made my comments based on that.
Slone does seem to have made the odd ill informed decision, based on the notorious instability of his symmetrical current mirror circuit. This is well documented so it is not unreasonable to reconsider his circuits.
I did say that the stopper resistors "could probably" be smaller and recommended that Cordell be consulted, not that they should be blindly reduced.
The compensation in the amp was new(ish) when Slone did it. That is just the point. He was not as well informed as we now are. The capacitor swap does not decrease stability and does improve performance. A pure win.
I recommended only to tune circuit values and not more complicated alterations.
Hence no TMC comments despite the impressive results and my interest in it.
The thread on Edmond Stuart's TMC front end is excellent and has some similarities to this circuit if you or the OP want to check it.

Best wishes
David

 

[CBS240]

Finally got time to update the design based on some of the feedback............

If you intend to use those lateral mosfets, and I assume you do because I see no thermal feedback compensation, I suggest you take another look at the datasheet and check the pin-out. Also if you use a gate Zobel filter like Bob shows, place a 'zero' from gate to drain, you can use a much smaller gate stopper and utilize the wide BW of the mosfet.


BTW one of your decoupling caps is shorted on the PCB drawing. Try making a leader wire of smaller width from the via and then switch to the larger width for routing. This way the end of the route doesn't bleed off the via or hole. Personally I don't like right angles for large tracks, plus the shortest path is the path of least resistance. If you are using the holes at the corner to ground through the chassis, this is probably not a good idea since the chassis should be grounded to saftey Earth gnd.

 

[homemodder]

The poster has decided to do his own board layout and therefore will need to test stability anyway so I made my comments based on that.
Slone does seem to have made the odd ill informed decision, based on the notorious instability of his symmetrical current mirror circuit. This is well documented so it is not unreasonable to reconsider his circuits.
I did say that the stopper resistors "could probably" be smaller and recommended that Cordell be consulted, not that they should be blindly reduced.
The compensation in the amp was new(ish) when Slone did it. That is just the point. He was not as well informed as we now are. The capacitor swap does not decrease stability and does improve performance. A pure win.
I recommended only to tune circuit values and not more complicated alterations.
Hence no TMC comments despite the impressive results and my interest in it.
The thread on Edmond Stuart's TMC front end is excellent and has some similarities to this circuit if you or the OP want to check it.

Best wishes
David

Slone did mess up with the symmetrical circuit but he never built or showed a amp to be constructed around that scheme. He simply showed in his book a way for higher linearity and loopgain using current mirrors. His build designs only used cascoding so I suspect he already new something was wrong.
The funny part of all this is that there are others I dont think need to be named that fell for the same mistake, just read previous editions of their books.

 

[nplez1]

If you intend to use those lateral mosfets, and I assume you do because I see no thermal feedback compensation, I suggest you take another look at the datasheet and check the pin-out. Also if you use a gate Zobel filter like Bob shows, place a 'zero' from gate to drain, you can use a much smaller gate stopper and utilize the wide BW of the mosfet.


BTW one of your decoupling caps is shorted on the PCB drawing. Try making a leader wire of smaller width from the via and then switch to the larger width for routing. This way the end of the route doesn't bleed off the via or hole. Personally I don't like right angles for large tracks, plus the shortest path is the path of least resistance. If you are using the holes at the corner to ground through the chassis, this is probably not a good idea since the chassis should be grounded to saftey Earth gnd.

I'll take a look at the datasheet to make sure my pinouts are correct. I did check briefly when I was designing the board, but maybe I missed something.

I caught the short after I ordered a proto board Fortunately, I was able to scrape off a few leads and add a jumper-wire to fix the problem. I've also removed the chassis ground connections. Thanks for the tip!

My current plan is to build the prototype and see how it works and then make some modifications to the design. We'll see how it goes...

 

[nplez1]

So, I checked the datasheet for the 2SK1058 and the 2SJ162 L-MOSFETS and it appears that the pinouts (on both) are 1 - Gate, 2 - Source, 3 - Drain. Looking at my design, this appears to be correct.

I know it is a bit confusing because I used BJTs in the schematic, but I didn't have a library with a TO-3P L-MOSFET. Am I missing something?