Help with changing opamps Tascam 3500 / with attached input schematics

[Matt Syson]

Hi
ANY new (faster) op amps will need 0u1 capacitors (or thereabouts) on the supply pins to prevent oscillation. Depending on physical layout it may need further measures.
Unless the electrolytic caps have been replaced the effect of doing so will be greater than swapping op amps.
The 470uF cap in the mic gain section should be much larger capacitance especially if you use a decent amount of gain. Many of the other 'coupling' caps are driving relatively light loads so are probably OK values as fitted.
The capacitors on the power rails are in series with the audio signal (although not apparent) so should also be replaced.
Matt S

[jrp]

The capacitors on the power rails are in series with the audio signal (although not apparent)

Could you explain this? I don´t think i understand...

[Matt Syson]

Hi
ALL current flows in a circuit, FACT.
For signal to 'flow' through a 'coupling capacitor' from one stage to the next, a circuit must exist and in an unbalanced audio situation it is via the combination of the power rails and the 'ground' connection (as a reference).
Each amplifier section takes DC from the supply rails and 'modulates' it's average 'quiescent' condition with your audio signal. It is assumed that the DC supply has negligible AC impedance, otherwise the amplifier will be working under variable conditions (distortion or instability). The supply rail capacitors help maintain this low impedance supply since it is common to 'isolate' stages using resistors or diodes and in any case you will have circuit trace resistance and inductance.
with the new super fast op amps the significant inductance of the supply and the gain of the amp can produce oscillation, hence the need for effective LOCAL capacitance (typically 0u1) but at audio frequencies an electrolytic is used to supply sufficient energy.
Matt S

[joe_masi]

Thanks for your answers. I have now changed micpre opamps in two channels and was really surprised. I only had time to do a quick A/B comparison with CD signals with the stock channels but the effect is really really subtle. I tried 2604, 2134 and 2132.
I will try some different signals like drum overheads the next days and hopefully there will be a greater audible effect.

I did the soldering without putting any 0u1 (0u1 means 0.1u, right?) caps in because I had no idea where to squeeze them in. What is to consider here? And is there a brand you would recommend or can I basically use any?
And how much larger should the 470uF cap in the mic gain section be?

While I was soldering a newbie question occured: What is so bad if an opamp oscillates somewhere in the MHz range?

The electrolytic caps have not been changed yet but now I think about doing that. But this sounds like soooo much work. Not that I am a lazy person but does the result outweigh the work?
And how much sense does it make to switch the summing opamps in the master section?
Best,
Joe

[Matt Syson]

Hi
I did the soldering without putting any 0u1 (0u1 means 0.1u, right?) Yes What is to consider here? And is there a brand you would recommend or can I basically use any? Almost any although some might suggest particular types.
And how much larger should the 470uF cap in the mic gain section be?
1,500 uF or 2,200 uF if it will fit.

While I was soldering a newbie question occured: What is so bad if an opamp oscillates somewhere in the MHz range?
Noise, distortion, possibility of interaction with others oscillating, chips burning, feeding 'RF' into other gear possibly causing failures.

The electrolytic caps have not been changed yet but now I think about doing that. But this sounds like soooo much work. Not that I am a lazy person but does the result outweigh the work?
Replacing caps is probably more important with fiddling with some op amps.
And how much sense does it make to switch the summing opamps in the master section?
Depends what is already in there and if they have extra support transistors.

Matt S

[joe_masi]

Noise, distortion and the chance of burning opamps? Yikes, you convinced me. Just to make it clear for me: Solder one 0.1u ceramic cap from pin 4 to a nearby ground and another one from pin 8 to a nearby ground? Is that right? I don´t want to destroy anything here.
Just wondering: What are the caps C135 - C140 doing? Somehow it looks like they are doing what you suggested. Or does local capacitance mean that they have to be as close as possible to the opamp pins?
In another thread I found this list:
Input A pcb-Change
Q101 and 102, to 2SC2546,
U101 to Lm6172,
C112,C 113 22pf,
C114 to 220uf with .01 poly in parellel,
C104,C105 to 47uf with .01 poly,
U102 to LT1358,
R111 to 243ohms,.01 poly across
C111,C112 to 22pf.
Kudos to Jim Williams for this one.

Except for the opamps does this make sense to you? I guess it does because the list is from Mr. Williams but this would be a good guidance for me.
Thanks again for you help I really appreciate it!
Joe

[joe_masi]

No one with a little hint here? Every tip is helpful.

[Matt Syson]

Hi
You should have finished the job by now, it only takes a few hours to do the first module (deep concentration) but the others get much faster as you are simply 'copying'.
Matt S
PS, sorry I missed the quotation marks on my previous post. Most was actually the quote and only a couple of lines from me, near the end.

[joe_masi]

Hi,
the bad thing is I haven´t started yet because I swapped back to the stock opamps after reading your post about burning etc because I still had some questions:

Just to make it clear for me: Solder one 0.1u ceramic cap from pin 4 to a nearby ground and another one from pin 8 to a nearby ground? Is that right? I don´t want to destroy anything here.
Just wondering: What are the caps C135 - C140 doing? Somehow it looks like they are doing what you suggested. Or does local capacitance mean that they have to be as close as possible to the opamp pins? On the pcb they are pretty close to the line opamp.

In another thread I found this list:
Input A pcb-Change
Q101 and 102, to 2SC2546,
U101 to Lm6172,
C112,C 113 22pf,
C114 to 220uf with .01 poly in parellel,
C104,C105 to 47uf with .01 poly,
U102 to LT1358,
R111 to 243ohms,.01 poly across
C111,C112 to 22pf.
Kudos to Jim Williams for this one.

Except for the opamps does this make sense to you? I guess it does because the list is from Mr. Williams and this would be a good guidance for me.
Thanks again for you help I really appreciate it!
Joe

[Matt Syson]

Hi
Yes do as the list suggests.
Note that the 0.1uF ceramics should go from plus supply (pin 7 or 8 depending if it is a single or dual op amp) to ground and from pin 4 and THE SAME GROUND POINT.
Some of the opamps are pretty expensive so after some recapping, try a couple of channels with new chips and see how you like it. You should of course make suitable 'mods' to master and monitoring modules otherwise you won't hear the extent of the improvements.
Matt S

[joe_masi]

Hi,
after finishing a job I now have the time to get back to the desk.
I decided to do the following:
- A pair of 0u1 ceramic caps from opamp pin 4 and 8 to a nearby ground, same ground point,
- Change C104 and C105 to a 47u electrolytic,
- Change C114 to a 220u electrolytic
- Replace C108 and C108 with new electrolytics, same value (Those were the ones in the powerrail you suggested changing, right?)

But now my last question to get started:
You suggested to change the C111 470u to a larger value like 1,500u but the list suggests to change it to 22p and C112 and C113 to 22p as well. So what to do?
Best,
joe

[Matt Syson]

Hi
On the basis that C111 is actually 470uF now, change it to 1500 uF or greater.
The list you had before is incorrect compared to the schematic you posted.
C111 has one end joining a 22 Ohm resistor and the other end on the gain pot.
Matt S

[joe_masi]

hi,
thanks for your answer! Never trust a list that someone got from someone and posted it somewhere elso on the internet...
So I will replace C112 and C113 with new ones but same values and C111 to 1,500u (if it fits, not much space here).
Great, I can´t wait to get started here..
Thanks again for your patience and answers!
Best,
joe

[jayadelic]

I just finished this mod on 2 channels & now the mic gain control doesn't work. Everything else is working properly. I used a MICA 22PF500VJ from mouser. Would changing the value to 1500v correct this problem? BTW the EQ sounds radically different now! It sounds great to my ears!

On another note, I swapped out the op amps to 2604's in 4 channels without doing the rest of the mods and to me they sound a lot better. It is subtle, but there seems to be a lot more clarity in the 250-125 range which has been my biggest problem with this mixer thus far. I don't usually track with the boards pres but occasionally I run out of other pres and have to. Things always get "murky" rather quickly. I haven't gotten to use it in a session yet but I believe this will help out immensely! No oscillation that I can hear.... yet..

[Matt Syson]

Hi
If you replace what was 470 uF with 22pF then of course it won't work.
You need to go up in value for which a multiplication of 3 or 4 is sufficient.
By fitting a 22pF you have DIVIDED the value by about 21 million.
Matt S
PS, you may not hear any oscillation as it will possibly be in the MegaHz region, depends what the actual circuits feel like.

[joe_masi]

Hi jayadelic,
murky is probably the perfect word. Especially if you record 16 tracks trough the pres and do the mixdown on the desk as well. I kinda like the sound though it´s a little too much.
The list I posted before was from the tascam forum and is not correct.
Have you already changed the C111 to 1,500u? If yes, how does it sound now?
It would be interesting for me to know what else you have done. Did you only change caps? Or transistors and resistors as well?
Did you change the opamps in your modded channels?
Have you changed anything in the master/monitor section?
I am really surprised that your EQ sounds different now because from my very basic electronics knowledge I would say all these mods only affect the mic pre.
Or did you change all three opamps in the input area?
best,
joe

[jayadelic]

I've never recorded 16 inputs with the desk pres, maybe 8 or 10 max. I have other pres but on a few projects I've ended up with like 22 inputs & just ran out. I generally only use 3 or 4 of the pres.

This is my first mod and to be honest I can barely even read a schematic I first looked at modding the master section but I couldn't find much info other than the star grounding mod. The master section had been newly recapped & the star grounding partially done when I purchased the board.

I don't use the busses so I didn't really see any need to re-chip them. The way I've been using it for mixing is to output pairs from my HD24XR into the line inputs on the channels then output the mains into a tape machine or back into the computer.

I did everything the Tascam forum suggested... I swapped the mic pre op amps with a 2604's and all the others with 5532's. I did change the transistors with 2N4401 becuase I was unable to fine the ones recomended on the list. The eq sounds much more "hi-fi". I guess would be the best way to describe it. That being said I never hated the eq's the way everybody else does. I actually like them better than some great plug in eq's I own. Obviously not surgical but more than useable IMHO.

I just ordered the 1500 uF last night so I'll keep ya posted when the arrive. I think there is plenty of room for 'em. I did have a helluva time getting some the polys to fit.

One more thing and I will finish my rambling, Matt thank you so very much! It is folks like yourself that make this place great! Cheers. Jason

[jayadelic]

So I swapped out C111 with a 1500uF and I am still having the same problem. The gain pot doesn't work. Everything else works fine. Any thoughts????

[Matt Syson]

Hi
If what WAS 470uF has now been replaced by 1500uF and it does not work then you have broken something, probably while unsoldering the old cap. Most likely to be a small crack in the copper print. In theory you could 'short out' the 470uF cap which would give you great LF performance BUT the pot would make nasty noises when turning it and depending on the 'offset' could limit headroom when set at high gain.
Matt S

[jayadelic]

I'm not exactly sure what you mean by "short out". Would this be achevied by just adding a wire jumper across it or would it have be grounded out to another point? Thanks!

[Matt Syson]

Hi
Yes a wire link across the capacitor terminals WOULD make a better sounding preamp BUT do NOT do it (except as an experiment) because the gain pot will be intolerably 'scratchy' (talking cones flopping out of your LF speakers) and probably premature clipping.
Matt S

[2N1305]

why such a high value? It's such a slow charge time, isn't that going to affect HF response? Typically these interstage DC block caps ar eno more than, say, 22uF... What's the story with this?

Thank you

(sorry to butt in, I just would like to increase my understanding)

2N1305

[Matt Syson]

Hi
The capacitor in question in THAT part of the circuit is working with low impedances (22 Ohms with gain pot at max gain (minimum rsistance)).
This means for a decent LF response that the reactance of the cap at say 20 Hz must be in the order of 1 Ohm or less.
It is assumed that the reactance of the cap is constant from the LF point of interest to several tens of KiloHz where it will rise slightly. You could argue about it's impedance variations at higher frequencies but this is a different story.
You are correct that interstage caps can be much smaller as they will usually be working into 5K ohm or often greater loads. It is usual to set a time constant between stages of say 1 Hz to minimise phase shift. Many older pieces of gear aimed for a higher frequency partly as there was not much interest as it could not be 'captured' and in some instances having a flat response down to DC (or almost) can be an embarassment. In this instance there WILL be differences in DC levels between the two transistor emitters which if allowed to extend to DC gain would lead to pretty big offsets. This can be removed by capacitor (here) or a DC servo circuit. As it is a 'cheap' mixer, the cap solution which is pretty reliable wins.
Matt S

[2N1305]

Thank you Matt. However, when you say a time constant of 1Hz, that doesn't really make sense to me, do you mean to say 1/1Hz, which is 1 second? That would be horrible as a time constant... I'm thinking no more than 1ms.

[Matt Syson]

Hi
I didn't express the time constant correctly but it should be a LOT lower than you may expect to minimise phase anomalies. DC coupling is seen as the 'holy grail' but gets expensive and complicated on large scale gear, typically requiring another 'op amp' for each 'node' that needs to be referenced.
Matt S

[2N1305]

Originally Posted by Matt Syson

Hi
I didn't express the time constant correctly but it should be a LOT lower than you may expect to minimise phase anomalies. DC coupling is seen as the 'holy grail' but gets expensive and complicated on large scale gear, typically requiring another 'op amp' for each 'node' that needs to be referenced.
Matt S

Yes I can believe that. That's why I love simple circuits. However our beloved tube amplifiers have coupliong caps everywhere so at some point you gotta get over it (not you, I'm just saying, people in general, people who are informed enough to know about it).

OK, sorry for making this a whole other thing!

Merry Christmas everyone! (Or happy Holidays)

[joe_masi]

hi there,
this has taken some time but now I am done:
I recapped and rechipped the whole master section, auxes, balancing pcb and 8 channels (I will do the other channels after a little break, enough soldering for now).
I put new Panasonic FM coupling caps with higher value all over the thing and a 1500u BP in the mic gain section. Interestingly if I turn the gain fully up my channel meter now stops at +6db and not +12db as it used to be.
Some channels now have 2604´s and some are just recapped. Now I am listening to a lot of stuff to decide which sound I like most.
Somehow the 2604 seems to be very accurate (but not sterile) but lack a little bass compared to the stock channels. Or am I just so used to how the console used to sound? Very hard to tell. Has anyone made the same experience?
I will order some more different opamps these days to have a little variety.
Any recommendations?

What things really cleared up: An 2132 as summing amp and 2134 in the monitor section. So much more definition and precision.

All in all I can say: AAAAMMMAAAAZZING!!
This sounds like a new console now (well, it almost is..)
Altogether it sounds like someone took away the carpet that was hanging between me and my speakers.
Thank you all so much again for your input!

One thing I about the local bypass: On the pcb was always a 0.01 u ceramic very close (about 0.5 cm with a direct connection to the supply pin and to ground) to each of the supply pins. This looked very much like a bypass cap to me. Is that possible that the circuit designer has already added them?
Nevertheless I have added 0.1 ceramics to be on the safe side.

And a last question: I know I should have done that before but do you have any suggestions for the psu? I have added the schematic down below. I was a little afraid that the new chips draw too much current but in the data sheets I found that an NJM4580 has an output current of +-50mA and an 2134 and 2604 has +-35mA. Is that the right value to compare and that means I am on the safe side?

All the best,
joe masi (very happy now)

[neil iceton]

Originally Posted by joe_masi

hi,
thanks for your answer! Never trust a list that someone got from someone and posted it somewhere elso on the internet...
So I will replace C112 and C113 with new ones but same values and C111 to 1,500u (if it fits, not much space here).
Great, I can´t wait to get started here..
Thanks again for your patience and answers!
Best,
joe

Hi Joe

I'm doing the same or similar mod here, i'm trying to source C112 and C113 ceramics and wondered what Dialectic Characteristic, and voltage you chose, for these caps, there seems to be a lot to choose from, and I dont want to get something that is not suitable!

Neil

[Matt Syson]

Hi
The 'reactance' of a 100pF cap is pretty small at audio frequencies. It would give a response 6dB down when the reactance is in the order of 13K so you can work it out. Consequently when used in this position the composition of these caps is not that critical as they stop the op amp oscillating and 'roll off gently' rather than 'peaking' at a few MHz.
Of course if you have changed the op amp you should also be reconsidering the value of these caps too. Fun isn't it?
Matt S