API 312 Thread!

[pmroz]

.

peter... or anyone..

are Jay's and Fabio's (Peter C layout) 1731 PCBs the same?

 

[clintrubber]

are Jay's and Fabio's (Peter C layout) 1731 PCBs the same?

I think they're in essence the same - at least the component placement is. W.r.t. the instability, you're thinking minor differences may come into play here ? If I understood it OK, the Jeffs simulations & the measurements both point in alike directions: not stable at lower gains. Unless Jeff has added certain PCB-parasitics it'd mean the design is likely the main source for the instability and not the PCB-layouts.

Would be interesting to hear from other 1731-users about their experiences and eventual 1731-tweaks.

Regards,

Peter

 

[louder]

alo
i have been using 1731 on loud sources,and did not hear any strange sound.when i have time,i will try to make more tests.
as sound concerns,i still find 1731 more edgy than api 2520 pre,although,1731 sounds more old(?).1731 marries very well with beyer m500(ribbon),but with gain all the way up,on vocals.well, but this is just me.
i followed peter c layout and component values.
best regards
pedro

 

[clintrubber]

Thanks Pedro.

I'm wondering, are all people for the AP*3*2-clone simply using the 1731 i.s.o. the 2*5*2*0 without changing any of the gain-setting values ?

More specific, are the 20k//120pF still used or a higher cap-value?

Regards,

Peter

 

[pmroz]

.

Peter, i have 2 channels Melcor. Tried with 120pf polsty and 220pf polypro. Went with 220pF, very slight difference. (Rob Flinn? i think tried 220pf first). First project, still waiting for scope to get here, so can't really test for oscillation yet.
Up until now, only ear testing. 1731s are nice-sounding, but other 2 channels 'BC' opamps were better for what i was recording.
Compared them both using hi-z ins and with in trafos. out trafos all 4804.
At the moment there is an Altec 1:4 and Cinemag 8PCCA on each.
Will try to A/B some things this week, and scope test as well.
Wish there was a way to route one mic to all 4 pres at once and record 4 tracks.

 

[Bauman]

I use the Melcor amps in 2520 place, just add another cap (100pF) from out to -in in the Melcor so when I plug in the API board I´ll have 220pF since there´s 120pF there for the 2520...

:guinness:
Fabio

 

[clintrubber]

Thanks all for the responses.

from Fabio:

I use the Melcor amps in 2520 place, just add another cap (100pF) from out to -in in the Melcor so when I plug in the API board I´ll have 220pF since there´s 120pF there for the 2520...

That's a nice approach. And who knows, it was just the way 1731's were used ? (with a higher R*C-product on their gainsetting-network). 20k&220pF, still an OK bandwidth. I'll add the cap & see how it behaves for low gain-settings.

BTW, there's a nice Jung-section on opamp-testing in the Jensen 990-paper.

Cheers,

Peter

 

[deveng]

That's a nice approach. And who knows, it was just the way 1731's were used ? (with a higher R*C-product on their gainsetting-network). 20k&220pF, still an OK bandwidth. I'll add the cap & see how it behaves for low gain-settings.

I've built a 2520 mic pre, using a 25.5k feedback resistor, (started with a) 62pf feedback cap, a 255 ohm series with a 25k reverse log gain pot. I found oscillation until I changed the feedback cap to 100pf. So, using a 100-120pf cap seems typical for many of our 2520 designs. The bandwidth is still good. This also showed up in simulations for specific versions. This one was solved with the feedback cap because of the location of the oscillation. I don't think we'll do that with the 1731.

I've not built nor bench tested a 1731 but I'm almost certain increasing the bypass cap value on the 1731 will only minimize the oscilllation problem. The feedback combination creates a rolloff quite a bit lower than the oscillation center. Yes it reduces the bump at 1.5Mhz but it's still an issue. If my schematic is correct for the 1731 we have an inherent design issue that needs to be dealt with. I believe increasing C3 to 470pf will help without causing any degradation to sound. I suggest this because it doesn't require any pcb modification. But, someone with a pcb will need to do it and test.

We could have at least 3 possibilities here. The schematic is incorrect, pcb layout issue, or an inherent design issue. To identify we'll need to scope these amps at low gain. Hopefully those with working designs are not misinterpreting the slight oscillation at low gains as part of the character of the 1731! If some indeed do have no oscillation then we've got a clue as to where to step next. I'll try to post my simulations in the next couple of days for all.

Regards,
Jeff

 

[deveng]

Sorry for mixing the 2*5*2*0 discussions in this thread with the 1731! I'll try to split the posts and still keep some sort of continuity (no pun intended) between related issues.

Regards,
Jeff

 

[clintrubber]

Cool, let's keep this moving !
And... wouldn't it be ultimate bliss if someone with some real original 1731s put a scope to it ? :grin:

And let's agree on the conditions.
Just a proposal:
* raw opamp, so no TXs in front or after
* some Hi-Z biasing-resistor to gnd (say 1MOhm)
* 22k & 22k gain-resistors (gain of 2),
* feedback-cap of certain value (under discussion),
* internal cap C3 of certain value (under discussion),
* supply +/- 15V,
* resistive load of ... Ohm
* capacitive load of ... nF

Jeff, could you update ? Let's use your simulation-conditions as much as possible.

Regards,

Peter

 

[pmroz]

.
Fabio, that's a great idea, hanging the extra 100pf there. Plug and play!
I was worried about too many feedback cap swaps pulling the pads.

Hopefully those with working designs are not misinterpreting the slight oscillation at low gains as part of the character of the 1731!

exactly right. I'm waiting for the scope for these (and 'BC's) before boxing them, to make sure they 'play nice'.
Couldn't say if they sound like original 1731. I have an old MAP5002 here, a little more 'shrill' on mid/high mids, with either 120 or 220pf. Both Fabio's opamp designs sound better than the 5002.

Peter, i hope i wasn't misleading about this in above post.
I'd love to test opamps only, but the best i can figure is on the pres with out trafo and hi-z in. I'll check that Jensen paper, thanks for link.

And let's agree on the conditions.
Just a proposal:
* raw opamp, so no TXs in front or after
* some Hi-Z biasing-resistor to gnd (say 1MOhm)
* 22k & 22k gain-resistors (gain of 2),
* feedback-cap of certain value (under discussion),
* internal cap C3 of certain value (under discussion),
* supply +/- 15V,
* resistive load of ... Ohm

A standardized test with optimal values on input and output is a great idea.
I'd feel better about the 1731s (and 2.5.2.0.s for that matter) if i could test this way. Scope should be here yesterday, and i can pad DAW output for signal, so hope to give a more scientific contribution soon.

Daveng, i'll also try a spare 1731 with 470pf for C3, and see what happens.

 

[clintrubber]

And maybe we'd specify the used transistors. I've used all -C types for Q1-Q4, -B for Q5,6 (that's of course BC550 for the NPNs, BC560 for the PNP). DMMs says the Q1 & Q2 I used are ~450.

But perhaps better skip this additional stuff for now, maybe the beta doesn't play that a big role here... (Want to put the circuit in a sim as well, like Jeff - and toy around with it)


Why I'm posting: I'd like to discuss that 1731-circuit a bit more, so that we're understanding what's actually going on there. Shall we move that to a separate thread or keep it here ?

Peter, i hope i wasn't misleading about this in above post.
I'd love to test opamps only, but the best i can figure is on the pres with out trafo and hi-z in. I'll check that Jensen paper, thanks for link.

No problem. And corrected for the TX-ratio, the load could be made identical, so in essence same setup. Note the 990/AES-paper can't be downloaded as far as I know but be requested @ the nice folks @ Jensen. They send it out free of charge.


Regards,

Peter

 

[CJ]

How low of gain settings are we talking about?
I have a couple of Melcors racked up if anybody wants some lab results. But I can not reaaly get inside the opamp without........dissecting it!!! :razz:

No, that's been done before.

Let me know what you want to check.

 

[pmroz]

.

Note the 990/AES-paper can't be downloaded as far as I know but be requested @ the nice folks @ Jensen. They send it out free of charge.

Well that explains it... I'll try an email. Thanks Peter.

Here all the Qs are 550C or 560C, with BD135/136 output.
Lost track of original opamp part #s. Got em somewhere...
I wonder what 'parts' daveng is using for sim?
Could be a variable...

This thread might be good to resurrect for 1731 talk.
http://www.groupdiy.com/index.php?topic=4796&highlight=sound+2%2A5%2A2%2A0
except for the name, but maybe daveng could change that....

CJ, oh transformer guru, do you know the Melcor 400-17?
looking for spec to see if these 3 are ok.

 

[deveng]

I think its a very good idea to standardize the test conditions as much as we can for testing both the 1731's and 2.5.20 designs. I suspect most of us are using both op-amps in the non-inverting configuration. So, here's my take on the conditions:

* raw opamp, so no TXs in front or after
This is good but may be difficult since most boards will be setup with
transformers. If we can test without all the better.

* some Hi-Z biasing-resistor to gnd (say 1MOhm)
If you're talking about the resistor from the non-inverting input to
ground I'd recommend using a 20k.

*22k & 22k gain-resistors (gain of 2)
On my test jig (I built one to test my 2.5.2.0's) I use a 12 position
switch. This allows repeatable gain settings from 2-100 (6-40db)
We should be testing at the full range of gains. Although, for starters
this would be fine to get over this initial oscillation problem. I agree
the most important now is low gain for the 1731 (gain of 2).


* feedback-cap of certain value (under discussion),
I suggest 120pf for the 2.5.2.0 and 200pf for the 1731s.

* internal cap C3 of certain value (under discussion),
Let's see if someone has success with my recommendation to increase
this to 470-680pf (on 1731's with oscillation problems). Also, lets see
if others who have built the 1731s have this issue as well. Otherwise
lets keep the schematic as original to begin testing.

* supply +/- 15V,
Yes

* resistive load of ... Ohm
Let's keep it at a easy load of 20k. I set my jig up to
switch to 600 ohms as well to test full drive capability. But for now to
keep is simple 20k would be a good place to start.

* capacitive load of ... nF
Let's wait on this one!

I should be able post my simulations in the next 2 days. I started with standard evaluation transistors but ended up using the BC550, BC560, BD135 and BD136 models. Simulations were essentially the same for frequency response. Both had the nasty top end on the 1731 with the bump and shelf at low gains.

We should be careful with simulation results. They can be good a tool in predicting circuit behaviour if they're setup right as seems to be the case with the bump in the 1731. But the bench is always the place to go for the real deal!

I'll follow where ever you decide to take this topic. At this point it seems to be jumping back and forth between both op-amp designs.

Regards,
Jeff

 

[clintrubber]

I posted:

I'd like to discuss that 1731-circuit a bit more, so that we're understanding what's actually going on there. Shall we move that to a separate thread or keep it here ?

but wasn't aware of this thread then:
http://www.groupdiy.com/index.php?topic=89


from CJ:

How low of gain settings are we talking about?

So far mostly about a gain of 2 (equal gain-setting resistors). Unity will make it of course 'even worse'.

Thanks for the responses, hope to RE more later today.

Peter

 

[clintrubber]

from CJ:

I have a couple of Melcors racked up if anybody wants some lab results.
Let me know what you want to check.

Thanks, that'd be nice. These are 'originals', right ? We were wondering whether the Melcor-DOA in its original dimensioning had ringing/osc as well at say gains of two. Let's hope that unlike the 2.5.20 there's only one original 1731-dimensioning :wink:

Regards,

Peter

 

[clintrubber]

BTW, everybody doing fine here with all this 1731-jumping between here &
http://www.groupdiy.com/index.php?topic=89&start=15 ?

I'm fine, thanks.

 

[Rob Flinn]

I haven`t had any problems with the Melcors since I added the 220pF.


The problem I`m having is with the relays & PSU. I swapped the 5v relays for 24v versions. Using individual dropper resistors for each relay I connected them to the 48v phantom line. Strange thing happens each time I energise a relay the 48v rail drops to the point where when all 4 relays are energised its down to about 24v ??? Obviously none of the relays energise at this voltage.

I thought it was beacause I was only feeding the TL783 with about 56v so I swapped the transformer for one that fed it 75v, & the same thing is happening.

I tried feeding the recifier straight from the transformer & NOT throught the decoupling 100uF caps & the same thing is happening.

I`ve double ..............quadruple checked the phantom side of the PSU & I can`t see any mistake.

Its probably an embarrassing mistake. Has anyone got any ideas ?

If I can`t sort this out I`m going to dump the relays & just use 3 pole switches.

 

[clintrubber]

from Jeff:

I think its a very good idea to standardize the test conditions as much as we can for testing both the 1731's and 2.5.20 designs. I suspect most of us are using both op-amps in the non-inverting configuration. So, here's my take on the conditions:

Thanks for the response.

* raw opamp, so no TXs in front or after
This is good but may be difficult since most boards will be setup with
transformers. If we can test without all the better.

As long as the loads are corrected by the squared TX-ratios it'll be OK, OK ? The 1MHz peak won't be too visible after the out-TX, but I figure it'd be fine to load the TX as usual after the TX and put the scope-probe before the output-TX, agreed ?

* some Hi-Z biasing-resistor to gnd (say 1MOhm)
If you're talking about the resistor from the non-inverting input to
ground I'd recommend using a 20k.

OK

*22k & 22k gain-resistors (gain of 2)
On my test jig (I built one to test my 2.5.2.0's) I use a 12 position
switch. This allows repeatable gain settings from 2-100 (6-40db)
We should be testing at the full range of gains. Although, for starters
this would be fine to get over this initial oscillation problem. I agree
the most important now is low gain for the 1731 (gain of 2).

You're right, we shouldn't stop once (if) the low-gain issue has been solved. Maybe the consequence of a certain fix is that at the highest gains there's not enough BW, so check indeed.

* feedback-cap of certain value (under discussion),
I suggest 120pf for the 2.5.2.0 and 200pf for the 1731s.

OK

* internal cap C3 of certain value (under discussion),
Let's see if someone has success with my recommendation to increase
this to 470-680pf (on 1731's with oscillation problems). Also, lets see
if others who have built the 1731s have this issue as well. Otherwise
lets keep the schematic as original to begin testing.

I hope to make a second one of the upcoming days and will put the other value in. Or I'll modify the first one.

* supply +/- 15V,
Yes

Anyone has some quiesc.current-consumption values ? Mine seem a bit low (8mA ?)

* resistive load of ... Ohm
Let's keep it at a easy load of 20k. I set my jig up to
switch to 600 ohms as well to test full drive capability. But for now to
keep is simple 20k would be a good place to start.

OK

* capacitive load of ... nF
Let's wait on this one!

OK

I should be able post my simulations in the next 2 days. I started with standard evaluation transistors but ended up using the BC550, BC560, BD135 and BD136 models. Simulations were essentially the same for frequency response. Both had the nasty top end on the 1731 with the bump and shelf at low gains.

That both a pity and promising - not too much sensitivity for the transistors, as should.

We should be careful with simulation results. They can be good a tool in predicting circuit behaviour if they're setup right as seems to be the case with the bump in the 1731. But the bench is always the place to go for the real deal!

Fully agreed. But your sims together with the scope-oscillations give faith in the sim-results.

I'll follow where ever you decide to take this topic. At this point it seems to be jumping back and forth between both op-amp designs.

... and an older Melcor-thread has joined I'm afraid (my fault, today :wink: ) must admit that's a bit less elegant. I've added crosslinks, let's hope that helps for now.


The big question remains of course (and as previously formulated): if the ringing is already in the original 1731, is it part of its sound ?
And I'm new to all these 312/2.5.20/1731-circuits, so I have no idea whether there was any 'original' use of the 1731-DOA in 312-like circuits. We're then more talking about the AM-27 then perhaps ?

Regards,

Peter