New Amplifier - ULD Extreme

Status
Not open for further replies.
AndrewT said:
V2.0c ULD shows a string of 4diodes and a variable resistor (100r) along with 6NJLs.

Could a fifth and sixth diode be brought into the string but appearing in parallel to the VR?

The one or two extra diodes can easily be shorted out with a link if the builder does not want to use them.
The 100r can be any value from 100r to 20k.

This would allow the tempco to be dialled in by reduce the effect of the fifth diode by bypassing it with the VR.

None of this would take up any extra PCB space. It would increase enormously the options open to the first few builders, who could then advise the later builders on which options seem to work best.

I am sorry if this seems a repeat of what I said before, but it's a no cost option that has been ignored.


Andrew
Another scenario could be to leave the VR as it is, but fitting the 6 diodes, where there is then the possibility of choosing 4 diodes with the highest forward voltages and then replacing 2 with links ?
Alex.
 
Andrew T

Actually if you look closely at the PCB that I showed in post #100 in the other thread all 6 diodes are brought out in the string. However I included 'shorting traces' that could be jumpered to shunt 2 of the 6 diodes. Does that work for you? Could my scheme be improved?

Actually it would probably make sense to make the adjustment pot in the diode string be the same value as the pot that is in series with the collectors of Q2 and Q4. Is 200 ohms a good value for both?
 
Andrew T,

I just re-read your post and realized that you were suggesting that the variable resistor be in parallel (not in series) with the 5th diode. I think that could be easily accomodated in the layout that I have posted. Is there any case where we would want the 6th diode in the string? If not the 'shorting traces' for the 6th diode could be made permanent on the PCB.
 
Hi all,

Carl, would it be possible to put three pads on the board edge next to the output devices to enable one to hang an extra output device on the heatsink external to the board. The extra diodes would not be used and the 0R22 could act as a jumper cable.
I still need the ability to go down to 2 Ohms reactive.

Terry
 
Hi Carl,

Nice drafting!

Yes, Terry wants it for 2R, so four pairs are his target.

Personally, to make this universally appealing, I reckon three pairs is tops. And three pairs at 56V rails should easily cope with 2R loads at moderate volume, and if reduced to 50V rails, would cope unconditionally.

It should be understood that 50V rails with a 2R load will do no less than 529 watts, assuming four volts loss from each rail and a stiff supply, a 46V peak signal output. 529 watts should be enough, even for rock 'n roll..... :clown: 😀

The vexed issue of the diodes comes up again and again. Since there is disbelief that they don't work as designed, best to leave them available, and bypassable if found inadequate.

Five diodes will give a reasonable bias voltage at 2.5mA. Four diodes and a pot in series will cope with the total VAS current, around 13mA, and it should be OK with 0.22R emitter resistors and generous heatsinking. Five diodes will require a bypassing resistor, around 210R, to carry most of the VAS current, leaving 2.5mA through the diode string. I have not tested this concept, so can't vouch for it. It's a risk......

Another option might be to leave the diode string, AND put in three pads strategically to fit a conventional Vbe multiplier.

Cheers,

Hugh
 
Thanks for that info. I need 300Watts into 2R. To reduce distortion to as low as possible, is it better to further reduce the rail voltage, how low can it go and what is the relationship if any between distortion and rail voltage.

Thanks

Terry
 
pheonix358 said:
Thanks for that info. I need 300Watts into 2R. To reduce distortion to as low as possible, is it better to further reduce the rail voltage, how low can it go and what is the relationship if any between distortion and rail voltage.

Thanks

Terry

According to Self there is a small relationship but nothing substantial. The higher rail volts will reduce the PN parasitic capacitances slightly.

If you have very reactive speakers you might be better off building a conjugate load matching network so that the amplifier seems a more resistive load.
 
AKSA said:
David,

Forr is always correct. He is 1 smart guy....

Y'know, I cannot see that the EF between LTP and VAS is necessary. I would prefer to see it AFTER the VAS.

Thoughts?

Hugh


Hugh / all,

I spent a bit of time simulating this amp the other day over lunch.

I tried:

1) - Pre VAS EF,
2) - Triple EF OP stage (or post vas as you say)
3) - Cascoded VAS
4) - Cascoded dif pair (following)
5) - Various amounts of dif pair degeneration
6) - Class A OP bias.
7) - compensation cap size
8) - a few other things

I haven't built it so I can't comment on any of the sonic changes
as applied to -this- amp.

I got the THD down to around 0.0005 / 20kHz / 200W / 4R.

As far as simulated results go:

1) Lowered distortion significantly
2) Made only minor improvements but if you wanted to approach
Halcro territory (-120dB) would probably be necessary
3) Also made minor improvements especially when used with 2)
4) Made virtually no improvements, but I think will resurface at
lower levels (near -120dB)
5) Made the most difference and this is plain to see, easiest way
to increase OLG. There is a point where stability becomes tricky.
I settled on around 33R depending on other changes.
6) Here is the big one. I think this is one of the most significant
improvements. The THD is not drastically reduced but the spectrum
is. When staying within class A, big reduction in 7/8/9/10 harmonics
etc. Also when not using triple EF OP stage often lower harmonics
actually worse.
7) The smallest compensation cap possible within the limitations
of stability and good step response gives lower HF distortion.

I haven't got time to get too involved in this as am pretty busy so
just blasted quickly through a bunch of iterations.

Obviously all these changes affect phase margin and hence stability
especially when using 1) 2) and 3) together.

If I get some more time I'll do a bit more but not sure when that
will be, if at all.

Anyway there's a few things to play with and consider.

cheers

Terry
 
Thanks Terry,

Only one surprise there for me; that EF post VAS made little difference.

Actually, I don't recommend a triple EF, but rather a buffer AFTER the VAS, then a double EF.

Your finding much appreciated. Degen and comp cap make huge differences. I've only ever found cascoding the input stage useful so you can use low voltage, quiet transistors in the LTP.

When Class A is invoked, something quite magical happens to the distortion profile. I've noticed it too in sims, even with diamond buffers. The problem is the death throes of the inactive output devices as they switch off.

Phase shift is a tricky one. I noticed radical phase shift with base stoppers, yet they are useful to improve stability. I strive to get phase shift below 5 degrees at 100KHz.

But omigosh, is Class A efficient or not?

It's not easy being green......

Hugh
 
Hugh, being green is only how you spin it. "I only use my class A in the cooler seasons. Then I use it as a low power electric radiant heater. But wait! Phone now and we will give you free, yes free the heaters ability to reproduce sound." See, class A = Green.

Can someone please tell me what happens to distortion as you lower the rails. Ta

Terry
 
pheonix358 said:
Hugh, being green is only how you spin it. "I only use my class A in the cooler seasons. Then I use it as a low power electric radiant heater. But wait! Phone now and we will give you free, yes free the heaters ability to reproduce sound." See, class A = Green.

Can someone please tell me what happens to distortion as you lower the rails. Ta

Terry


Or run a switchable bias......
Foe when feeling guilty.
 
Status
Not open for further replies.