JLH 80w mosfet power amplifier - modifying it

[Kevinbd]

Hello,

I am currently trying to modify a JLH mosfet amplfier. This was designed back in '82.

What I am trying to do is replace Q10 which is a VN1210M (1w) vas fet with a higher power VN1210D device. (45w). Obviously the VAS CCS and vbe transistors will have to be upgraded and probably heatsinked. Enno Borberly in his earlier designs such as the DC100, Servo 100 etc recommend driving lateral Mosfets with about 30-40mA per pair. This was because of the capacitance, apparently this improves HF THD and also the slew rate. Another thing that JLH does in his design is uses different gate resistors (presumably to match the bandwidths closer). But I would prefer to use equilisation capacitors over the N Chl mosfets as I do not think the slew rates could be symmetrical ?? (not sure on that one)

My questions is, can I replace the VN1210M with a higher power VN1210D and will this affect the compensation of the amp ie. need different compensation capacitors etc. I am not good enough to adjust this to be honest.

I have included a link to a datasheet which shows the devices in question. One minor problem is that it does not show the VN1210M by name.

Just wondered if anyone had any old supertex databooks with these devices in particular. Think I am alright though.

http://pdf.alldatasheet.com/datashee...X/VN1210D.html

Any help appreciatted

Kevin

[Kevinbd]

Hello,

Oops I forgot to mention that Q10 Fet in the standard circuit passes 10mA - hence the question.

[lineup]

Thanks for schematic Kevin
We are discussing JLH 80 Watt MOSFET in another topic
and did not know where to find a schematic for this amp.

Suddenly you posted this thread!
Your schematic was really 'heaven sent'.

see this topic:
Effect of Power Supply on Soundstage Width

Another guy, foxyb, is modifying his JLH 80.

[Geoff]

Originally posted by Kevinbd
I am currently trying to modify a JLH mosfet amplfier. This was designed back in '82.

Just to avoid any possible confusion for people reading this thread, the schematic you have posted is not the circuit that was published in Wireless World in August 1982 (80-100W MOSFET Audio Amplifier). It is the schematic for the later 'Audio Design Amplifier' which appeared in the July 1984 edition of ETI.

[Bobken]

Hi Kevin,

I have the Supertex Databook dated 1991, but the 'M' suffix VN1210 device is not shown in this either!
However, from the book it is clear that with these various VN1210 Supertex family devices, provided there is merely one letter in the suffix (i.e. 'M', or 'L', or 'D') there are no differences as far as their electrical parameters go. The differences here are merely their packages and therefore their power ratings/heat dissipations etc.

Where there is a number added to the suffix as well as the following letter (i.e. VN1210N2, VN1210N5 etc.) this no longer holds true, and their electrical parameters are different.

As the 'D' suffix devices you mention are T0-220 packages, it would seem that they will be fine in the intended application.
The original 'M' suffix Mosfets (I have some to hand) are merely the size of a TO-92 with a small metal tab on top (I have not come across these devices before, nor since, so I don't know what the packages are called!) but their dissipation will be much less than a T0-220.

You mention the gate-stopper resistors, R18/19 & R21/22, and you are correct in your assumption that R18/19 are higher values than R21/22 to 'balance' the different capacitances of the output devices, to bring the HF roll-off points nearer to each other.

You may also be interested to know that the designer later recommended that these be increased to 330R for R18/19, and 270R for R21/22.

I hope this helps.

[amplifierguru]

Why does this cct have a Vbe multiplier for Tcomp of lateral MOSFETs? With Re?

This would be excessive (doubly) even for verticals.

Cheers,
Greg

[srh]

Hi Greg

IIRC the vbe multiplier is not mounted on the o/p heatsinks, it is not intended for Tcomp, but to generate a constant bias voltage.

Cheers,
Steve.

[Bobken]

srh is quite right here.

Q8 is mounted on the circuit board and is a simple 'amplified dode'
used to control bias, with no intention of any compensation effects.

[Fanuc]

Hello,

Lineup quoted the following on another post about soundstage width.

"The gain is slightly less than (150k / ( 0.82k + 2.2k // 0.5k ))+1
Would make <123.2 or <41.8 dB, because impedance of C7, 9uF4 is not zero."

My question is can the NFB impedance resistors be scaled down. The reason for this is to reduce the noise and DC offset. I personally do not like the cap & trimmer (RV2 & C8) over the LTP emitters. Plus there is degeneration resistors to consider as an improvement.

Could I change R1, R14 to 75K (to start off with) and change R8 to 410R ?

Would polyprop cap C7 have to be scaled up to 20uF also ? Would like to keep the polyprop cap if possible.

I figured after reading Doug Self's book regarding degeneration that re = 25/0.25 = 100. So if the current is doubled in the tail source I could add 50R or 47R resistors. (I think the pot has to go with degen resistors ? not sure

The only hole in the head to all this is whether stability is radically altered!. Here is a link that discusses the gain being to high for modern sources.

http://www.diyaudio.com/forums/showt...014#post788014

Best Regards

[Bobken]

Hi Fanuc,

The answer to your first question is certainly 'Yes', you can scale the resistors and caps within reason.

In fact the only reason the designer (JLH) used much higher resistors here than previously, was that when he visited me I had demonstrated the sonic improvements to be had when using film caps for DC feedback blocking, in place of original electrolytics.

I had previously tried the much higher than usual 150k for the feedback R14 (which then needed R1 to be increased accordingly to avoid any offset) together with a single 4k5 (in place of the entire 'balance' arrangement) to ground, to enable a 10uF film cap to be used in series here.
I didn't care for this 'balance' arrangement, at least not what it did to the sound, mainly because of the variable pot used here in a very critical part of the circuit as far as 'sonics' are concerned.

The same goes for any electrolytic cap (which I was then aware of) because all of those which I had tried sounded awfull compared with a wire bypass I had also previously tried.

You are also correct in your assumption that if the feedback resistors are decreased in value, to preserve the intended lower cut-off here, you will need to increase the cap's value proportionately to achieve a similar time constant.

This calculation is made using the value of the smaller resistor alone (in my case the 4k5 to ground) and the chosen cap, and not, of course, the higher value feedback R14 which does not affect this issue.

Wishing to achieve a -3dB (lower) cut-off which I thought was suitable here, using a 4k5 resistor which I had to hand, gave me a calculated result in the region of 3 to 4 Hertz (IIRC) when used in series with a 10 uF film cap which I had obtained for this purpose.

I also wished to achieve an overall gain of approx. 35x, so with using 4k5 for the smaller resistor, this required approx 150k for the feedback R14. (150 + 4.5/ 4.5 = 34.33x) None of these choices is really very critical in practice.

There was a potential for slightly more noise as a result of the increase in resistances, but I chose some better quality components here to help offset this, and the overall sonic improvement was sufficiently impressive to JLH for him to adopt this arrangement in future designs.

Incidentally, the output DC offset is mainly governed by the matching of gains of the input LTP, and the closeness in matching of the total resistances from Q1's base to ground (i.e. R1 + R2) and the value of the larger feedback resistor, R14. If these gains and total resistances are exactly equal, theoretically the offset should be approx. zero, and in my case, I ended up with far less than 10mV in both channels.

I also got rid of the offset trimmer and its bypass cap which you mention, and used 47R 'fixed' emitter resistors here, although I also made some other changes in this area.

I have never experienced any stability issues with this design, in spite of many experiments made over several years, so, within reason, you should be fine.

I hope this helps.