Melcor Deconstructed

[rafafredd]

well, on the melcor original boards the layout was quite a mess...

All compensation caps was conected with isolated wire from one end of the board to the other.

 

[deveng]

The internal feedback cap (C2) on this schematic has been driving me nuts for weeks! My simulations show some severe peaking at high frequencies that is especialy bad at low gains. It appears this design does not respond well to internal compensation either by local (Miller) or global feedback. Changing the cap from 47pf to 10pf did not solve the problem but made it worse. I've done some manipulations with the circuit and finally found a combination that simulates well (so far). First, you can leave the cap C2 at 47pf but do move the connection to the positive input terminal as has been suggested. This reduces peaking but it is still there. Then I found that by adding a single 39k resistor between the collector of Q3 and to the base of Q5 the response curves are much better, even at low gains. It still does not look great in an inverting configuration. If you're using it as a non inverting amp it should work well (gains of 6db or above).

I don't have a pcb to test this on so if anyone would like to try this mod, it would be interesting to see if it works as good as it simulates.

I'll keep at it!

regards,
Jeff

 

[bcarso]

I'm too lazy to hunt for it, but is this the amp that PRR suggested wasn't necessarily made for low gains? Maybe not.

I spent a fair amount of time in sims, but most were before it was revealed that the single bias diode was at least a drafting error. I forget where I wound up, but it was with a fairly different set of compensation values for optimal transient response.

 

[deveng]

bcarso wrote:

I'm too lazy to hunt for it, but is this the amp that PRR suggested wasn't necessarily made for low gains? Maybe not.

I spent a fair amount of time in sims, but most were before it was revealed that the single bias diode was at least a drafting error. I forget where I wound up, but it was with a fairly different set of compensation values for optimal transient response.

Yes it does (did) seem to not to be happy at gains less than 5-10dB. Inherent design problems it seems (or schematic errors). All my attempts to use compensation for the high frequency peaking just resulted in more problems including oscillations as it was overdriven. It seems to me it comes down to the connection between Q3 and Q5. It makes sense that
A*P*I redesigned this thing. It's not the greatest to start with and has lots of issues. Adding the series 39k resistor seems to be a huge leap but I won't really know for sure until its bench tested. I still think this design can be improved within reason and hopefully maintain is character.

Regards,
Jeff

 

[deveng]

Sorry, somehow this message got double posted!

Regards,
Jeff

 

[Flatpicker]

Hi Jeff,
Which transistors are you using to model this circuit? I assume the BC550 and BC560?

Thanks for all the hard work. :thumb:

 

[deveng]

Flatpicker,

The latest sims were the BC550C and BC560C with BD135/136 outputs. I've also done sims with TIS98, TIS97, 2N5087 etc. The transistor doesn't appear to change things much. Biasing resistors need to be adjusted to get the exact same currents but other than that, the sim results are very similiar.

Regards,
Jeff

 

[Bauman]

I still think this design can be improved within reason and hopefully maintain is character.

This is important cos it does sound interesting to me...

If we can list the proposed improvents I can test it. LEt me try to sum up:

1) C2 = 47pF, and goes to Q4 (B) instead of Q3 (b)
2) 39k resistor between Q3 (C) and Q5 (B)
3) dual D2 (thnx Gary)


what else ?

 

[gar381]

Hi Fabio

Don't forget the dual D2.

GARY

 

[Flatpicker]

[quote author="deveng"]...First, you can leave the cap C2 at 47pf but do move the connection to the positive input terminal as has been suggested.[/quote]I understood rafafredd to say that it is connected to the ?-? input (base of Q2), not the ?+? input.

I'm just thinking out loud here comparing the 1731 with the 2520?
On the 2520, a similar positioned cap connects to the base of Q3. Connecting it to the base of Q2 would reverse the phase of what the cap feeds back into Q3.

 

[gar381]

Hi all

Has anyone had a chance to try and listen to any of the changes yet :?: :?:

Would love to hear anyone's thoughts. :grin:

GARY

 

[mhartung]

The boards will not be trashed by this. We can connect one leg of C2 to the jumper point and make the jumper on the solder side..

This is really easy as the Millmax pin of the (-) input is only 5mm away and can easily stretch the leg there. Luckily, the head of the pin sticks out of the top of the PCB and can easily be soldered to.

hope this helps

mike

 

[gar381]

Bump

Fabio or anyone

Does anyone have anything new to add on the 1731 changes? :?
How do these changes change the sound?? I will be getting back
to playing with mine soon and wonder if I should make these mods.

GARY

 

[peterc]

Something I have found, is that the double diode gives the opamp better stability in this way (used in a 312 circuit) :

With a single diode, there has to be at least a gain of 3 or else the cct would be unstable.

With a double diode, the 312 cct is stable for a gain of 2, but is still not stable with a gain of 1.

No other mods tried though.


Peter

 

[rafafredd]

I can only confirm what I said before. 10pF goes to the base of Q2, directly to the inverting input... I thought it was really strange, but I do reacll Bcarso saying it was OK, in a post, possibly in this thread. I´ve checked the original PCB many many times for this, and ther is actually a component that goes directly to the inverting input. Based on my notes, and on my memory, it was really that 10pF. I did lots of errorr because of inexperience and because of vintage transistors strange pinouts comparing to the datasheets I´ve found on the web, but I am SURE that this 10pF compensation cap was connected there.

Has anyone tried this mod?

 

[Jonkan]

I tried it, and it made my transients sound strange when i played my ac. guitar hard.

Could be i ****** up something else in the process, im gonna compare two channels soon, one "unmodded", and one where one side of c2 goes directly to the - leg.

Also tried the 39k resistor between q3 collector and q5 base. Made the whole thing behave even more strange when handed strong transients

Perhaps it was my mics? (modded mxl 603s, Thank you Gus)

Though that also seems strange since they have been working fine lately.

I also replaced C1 with a 1000pf Polystyrene cap (instead of a 680pf polyester film cap). Could that have affected anything?
/ jonas

 

[Flatpicker]

[quote author="Jonkan"]I tried it, and it made my transients sound strange when i played my ac. guitar hard...[/quote]That sounds like what the 603 mod does to the mic - It makes it smoother on some things, but kills the acoustic guitar transients IMO. Have you tried any other preamp configuration since the mic has been modified?

 

[Jonkan]

I tried the same preamp with the 603s before i modded the opamp. With c1=680pf.

Cant say that i reacted the same way about the sound back then.

Also tried my mindprint di-port pres.
Still sounded good, but it was a while ago so i cant be 100% sure about all of this.

Could it be the polystyrene cap in the melcor opamp that has caused this? Should i change back to 1000pf COG/NPO?
/Jonas

 

[clintrubber]

Could it be the polystyrene cap in the melcor opamp that has caused this? Should i change back to 1000pf COG/NPO?

IIRIC someone hadn't noticed much influence of polystyrene (was it Peter C ?)
Could imagine the value has more influence than the type of cap - as long as they're 'blamefee' types, like polystyrene or COG/NPO.
If stability is marginal then the value could show up more in listening tests.

But what are you after, the most neutral possible result from this circuit or a reproduction of the original ?
Can imagine the original didn't use the best type of components possible (Rafa might know more), so if you want it to sound like that maybe you should actually use more crappy types of components.

Bye,

Peter

 

[Jonkan]

I just had some 1000pf polystyrenes lying around that i wanted to try. No calculated thougts about it.

I have never heard the original, so its kinda hard to compare. I just want a preamp that sounds great.

Im gonna mess around some more with this, try a few different mics, opamp configs, etc and see what gives.

/Jonas