Melcor Deconstructed

[clintrubber]

Was toying around with the circuit in a simulator and in the meantime all these fine (re-)discoveries turned up. Nice.

What I found of the 'old' circuit:
(FWIW, but since I already typed it down offline here it is anyway)

Some sim-toying around, with totally different transistors, so all this VERY FWIW.

Conditions: +/- 15V, internal caps initially held at 630p/10p/10p and feedback-cap 200pF.
Gainsetting-resistor 220...22k & feedback-resistor 22k. Load-resistor 20k.


* DC: with a single diode for D2 the output devices Q6,7 don't conduct

* an AC-sim comparison showed the influence of the single vs double diode for D2:
while there's some peaking @ 2MHz for the double-D-situation for gains of 20dB and less,
it's not awful.
At 6dB gain it's only a few dB.
For the single-D-situation the peaking is more severe and decreases in freq
for higher gains.
Like:
"7dB @ 1.2MHz for 6dB gain" down to "2dB peaking at 200kHz for 40dB gain".


Identical D-vs-2D-comparison for 6dB gain with C3 boosted to 470pF as per Jeffs suggestion:

For the single-D-situation the peaking is some 8dB @ below 200kHz

For the double-D-situation there's even some more (~10dB) peaking, @ 350kHz

More detailed analysis will be of little use, my transistors are not even close etc.
I might have gone already too far with this present sim-setup.
I only continued with it after seeing the peaking was in the same ballpark as the previous simulations of the fine people here.

Good to hear the double-diode is being confirmed, thanks Rafael.
Transient-sims show that the THD-residual is now virtually absent, i.s.o. nasty for the 1-D-situation (Bcarso already commented like this). And the output-devices are now actually conducting i.s.o. wildly switching around.

Those were the only changes. Let me know if you need me to make another schematic with the changes I´ve listed.

Yes please, would be great - that'll bring us all on the same page again.

Haven't checked yet, but it sounds like the corrections can easily be added to the already built PCBs.

Regards,

Peter

 

[Flatpicker]

Thanks for all the work, rafafredd. I'll update my schematic tonight.

 

[bcarso]

Peter, confirmed---the addition of the diode cleans up all the crossover distortion. I went ahead and moved the connection of C2 as well. I now see mostly second and third, and things are the usual simulator-unbelievable at 0.4 ppm THD, since thermals, input C variations with voltage, and component mismatching are neglected.

Still trying to figure out why R8 is needed---current speculation is that is has something to do with avoiding a latchup condition.

What are/were the rails supposed to be for this part?


Brad

 

[clintrubber]

from Brad:

since thermals, input C variations with voltage, and component mismatching are neglected.

You're right, sims are often too friendly. Talking about thermal, we won't have the thermal feedback of output-devices to the diff-pair here, or even if potted, then at least way less than in an integrated (interrogated ? :wink: )circuit. Could model that thermal coupling of course in sims, but it won't be needed here I expect. But I must say I'm not at all familiar with the thermal-speed of those epoxies.

What are/were the rails supposed to be for this part?

Fabio commented on this one day, I thought he said +/-15 to +/-20. But it'll depend on the used transistors of course, if we wanted we could go further. But since there'll be an up-transforming TX following in many cases, I guess +/-20V is more than enough.

Bye,
Peter

 

[PRR]

> PRR had a look at this and redrawn it with the right position of the transistors and we ended with a working schematric.

The connection of the 18K is probably wrong.
Capacitors not shown: we were still trying to get the transistors the right way around, as shown in this scrap:

> What are/were the rails supposed to be for this part?

I had an impression that it was +/-24V, but can't say where I got that idea. And I see in the plan above I assumed +/-15V.....

On the Flickinger, it was very clear. Less than +/-18V gave zero current, over +/-27V the current ran-away, and only +/-23V to +/-25V really looked like "reasonable" currents. In SPICE, plot a couple key currents while you sweep the power supplies, and look for the happy-zone. This amp may be far less critical than the Flickinger.

 

[pmroz]

.
About rated voltage: the MAP AM27
http://img.photobucket.com/albums/v357/plushmore/QE712/am_27.jpg

rates +/-15v. I know this is not Melcor (which would be AML27) but according to Paul Wolf interview (link in AM27 thread), MAP were early Melcor knock-offs and used Melcor trafos. Would it be reasonable to guess the ratings are similar?

 

[peterc]

As far as I can remember, I did not make space for the second diode, it was left to the constructor to do a double 1N4148 diode.

Peter

 

[Flatpicker]

Seems like now I remember there being an extra diode too. Don't know why I didn't fix the schematic. Oh well, it's fixed now.

 

[Flatpicker]

[quote author="clintrubber"]from Flatpicker:

A blast from the past!.

Funny, this circuit looks so simple now...

Ehh, OK :wink:...[/quote]Hey, just because it looks simple still doesn't mean I completely understand it. :wink:

 

[clintrubber]

Hey, just because it looks simple still doesn't mean I completely understand it.

Agreed ! :thumb:

But hmm, we don't completely understand women - does that mean they're simple ? :roll:

 

[clintrubber]

from Flatpicker:

Seems like now I remember there being an extra diode too. Don't know why I didn't fix the schematic. Oh well, it's fixed now.

Thanks for the corrections. Wait, rereading Rafaels post, how about the left plate of C2 ?
And those 10nF decoupling, oh well,. But (without having resoldered it yet) that C2-connection looks relevant. Perhaps...

Thanks,

Peter

 

[clintrubber]

Hi,

I know this is not Melcor (which would be AML27)

FWIW, the middle pic does mention 1731A.

 

[pmroz]

.
... i miss a lot. MAP must not have made their own opamps yet.
Might try to open MAP 5002, see if it's any relation.

so does +/- 15v sound right for testing? I couldn't tell sound difference from 15-18v...
BTW 1731 does sound better with D2=2x1N4148 than D1=2x4148. :grin:

 

[Flatpicker]

[quote author="clintrubber"]...Wait, rereading Rafaels post, how about the left plate of C2 ? And those 10nF decoupling...[/quote]Remember, my schemo is for the boards that were circulating at the time, which used the 10nF decoupling caps. Not sure about C2. I'll check it on one of my boards when I get time.

 

[clintrubber]

All fine, thanks :thumb:

 

[Jonkan]

Hello everybody!
Been trying to follow this thread. Cant say that i really understand everything :green: . But basically you're saying that there needs to be 2x1N4148 in the D2 spot on the pcb? Have i understood this correctly?

How do i connect these, serial or parallel?

Whas there any other things, caps etc, that we all can agree on needs to be changed?

thanks!
/Jonas

 

[clintrubber]

Hi,

See the updated schematic from Flatpicker on the first page of this thread.

Bye

 

[Jonkan]

Thanks!

sorry to bother with my questions again, but which caps are the 10nf decoupling caps? I can only find 10pf (c2,c3) and 100nf (c4, c5) caps in the schematics (except for C1 that is..). Which ones are you discussing?

/Jonas

 

[clintrubber]

Yes right, these are 'under discussion', as is the connection of the left plate of C2.

But the essence is the added diode, I guess it all will greatly benefit from that and those additional things are of lesser influence. And you already mentioned C1, it should be 680pF, as indicated in the updated schematic.

Bye,
Peter

 

[Jonkan]

Thanks! i got that much from the schematics.

I just wanted to know if you accidently had typed 10nf instead of 100nf in the discussion, since there seems to be no 10nf caps in the schematic from flatpicker. There are however 100nf ones. Got me a bit confused :green:
/Jonas