|
|
Post by matt@IAA on Dec 5, 2017 22:09:49 GMT -6
Edit: Jump down a bit for the pics of the board and such.
Alright. I have a SoundWorkshop 1280b-8EQ that I'm recapping among other things (correcting it for +4 dBu monitoring per the manual, rewiring to pin 2 hot, and so on).
These originally have unbalanced line ins and outs. They use a 4741 IC opamp, and every output source impedance is 47 ohms. It's also designed for -10dBV (super specific design to a Tascam 8 track tape machine) but the I/O cards are designed to handle +4 level.
I've seen people put 10k:10k on the inputs and 600:600 on the outputs. I have a hard time seeing that this is worth the cost, but I'm toying with the idea of putting a pair of output transformers on the mix bus, because why not?
I've been mulling this over all day and I've only managed to confuse myself.
Here's my questions. (Later answers edited in).
Transformers are labeled for their impedance numbers based on bandwidth spec, right? They don't themselves impart an impedance load to the circuit. So, wouldn't the ideal 1:1 transformer for this application be one spec'd for 47 ohm:47 ohm, if such a thing existed?
Answer: Transformers are optimized for certain impedances since their inductance creates an RL highpass filter with the source resistance. So the inductance of the transformer (which is loosely related to the DC resistance) is adjusted based on the expected source resistance. Both are 1:1, so they would have the same reflected impedance from whatever load you put on, but a different frequency response.
Next...if this circuit is designed for a 47 ohm source impedance, and probably driving a 10k ohm load impedance, what would happen if you put a 1:2 output transformer on it? Am I right in thinking that your 10k ohm impedance on the secondary becomes a 2.5k ohm impedance on the primary side? So now the opamp "sees" 47 ohms to 2.5k ohms instead of 47 ohms to 10k ohms? I'm thinking this increases the output current on the opamp by 2x.
Answer: Transformer distortion comes from source impedance vs a nonlinear load. Low source impedance means you have lower distortion, so directly coupling an output transformer with an opamp would make the transformer happiest, but not necessarily the opamp. Reflected impedance goes with the square of the turns ratio; current and voltage are proportional to the turns ratio.
Specifically looking at an API 2623 style output transformer, 75 ohm primary to 75 (1:1) or 300 (1:2) secondary. 1:1 would seem to be a transformer with an appropriate bandwidth for the impedance it's being presented on the primary side (47 vs 75 ohm) and a nominal output source impedance change from 47 to 75 ohms. On the other hand, 1:2 would add 6 dB of gain to the 26 dB gain available on the bus, bringing the nominal output up from -2 dBm to +4 dBm...But would it hurt the op amps?
Answer: Basically the 2623 at 1:1 is probably fine but maybe not a good idea to do 1:2. Impedance ratio is the square of the turns ratio, so 1:2 turns ratio is 1:4 impedance ratio. With 1:1 and a 600 ohm load the op amp would "see" the 600 ohm load plus the impedance load from the primary on the transformer plus the existing 47 ohm load resistor. When wired 1:2 the op amp "sees" the 600 ohm load as a 150 ohm load plus the impedance load from the primary and the 47 ohm load resistor, so the load impedance drops a lot. If you look at the 4741 datasheet, the max output voltage swing starts to drop dramatically after 1k load resistance. At 675 its about 18V. At 250 its around 12V p-p. Sending +/-15V the 4741 output is 15mA ... trying to drive a 15V peak into 250 ohms requires 60 mA! A bit too much to ask.
Thanks in advance!
|
|
|
|
Post by matt@IAA on Dec 6, 2017 14:31:24 GMT -6
Ok... so Dave at Cinemag squared me away. I was on the right track. All of my questions are answered now. (And yeah, this would be similar to similar to shorting one opamp on the chip to ground, which the datasheet says it can be done indefinitely but doesn't seem to be a good idea, so I'm not going to try it.)
|
|
|
|
Post by johneppstein on Dec 7, 2017 13:45:51 GMT -6
You might think about replacing the 4741 opamps with something more up to date, they're pretty archaic.
|
|
|
|
Post by matt@IAA on Dec 7, 2017 22:58:40 GMT -6
Funky mojo nasty is ok with me.  |
|
|
|
Post by matt@IAA on Dec 9, 2017 11:00:27 GMT -6
Decided to turn this into a rehab journal -- hope that's ok. My work list: 1. Upgrade PSU capacitors from 2700 uf 35V to 4700 uf 50V. 2. Control room card: - Add a bypass MKP-2 0.1 uf – 250V film capacitor from the (+) power pin to (–) power pin (two middle pins, 4 and 11) on the 4741. This is done underneath on the solder side of the PCB, laid flat.
- Change resistors R8 and R9 to 150k Ohms to accommodate +4 dB level monitoring per the manual.
- Replace electrolytic capacitors. Increase all capacitance by 10x except for C3 and C7 which are timing caps for the solo circuit
| Control Room Card | Existing: |
| Replace with: | | | Cap | uf | V | uf | V | | C1 | 22 | 16 | 220 | 35 | | C2 | 22 | 16 | 220 | 35 | | C3 | 1 | 50 | 1 | 62 | | C5 | 10 | 25 | 100 | 35 | | C6 | 10 | 25 | 100 | 35 | | C7 | 1 | 50 | 1 | 62 |
3. I/O cards: - Add a bypass MKP-2 0.1 uf – 250V film capacitor from the (+) power pin to (–) power pins.
- Confirm installation of RP 6.81k1% resistors (2 per channel) for phantom power per the schematics.
- On channels 1-8 change resistor R37 to 47k Ohms to accommodate +4 dB level monitoring per the manual.
- Replace electrolytic capacitors (*C10 is a radial all others axial). Increase all capacitance by 10x
| I/O Card | Existing: |
| Replace with: | | | Cap | uf | V | uf | V | | C3 | 10 | 25 | 100 | 35 | | C9 | 22 | 16 | 220 | 35 | | C10* | 10 | 25 | 100 | 35 | | C11 | 22 | 16 | 220 | 35 | | C12 | 10 | 25 | 100 | 35 | | C13 | 10 | 25 | 100 | 35 | | C14 | 22 | 16 | 220 | 35 | | C19 | 22 | 16 | 220 | 35 | | C20 | 22 | 16 | 220 | 35 |
4. XLR connections - Switch all XLR connections from pin 3 hot to pin 2 not 5. Phantom power is wired pin hot, sleeve to pin 1. I need to find and fit linear 48v 1/2A power supply and barrel pin connector. 6. I decided to put in a pair of output transformers on the mix bus. This little board is interesting, in that it's highly specific to an 8 track machine. So there's 8 busses, and then a 2 mix. But, during final mixdown you route everything to bus 7/8, and it's a common circuit with the 2 track out. In this case, that's great for me. I have one stereo bus with two outs, so I can make one colored and the other not. After talking to Dave at Cinemag I'm adding two CMOQ-2S, which he recommends at 1:1 with a 1.8k ohm termination. I'm going to try to fit them in the case and the stereo outs it with 1/4 TRS instead of the existing RCAs. |
|
|
|
Post by matt@IAA on Dec 9, 2017 11:25:00 GMT -6
First things first. Get all those knobs off.   Exposed, you can see how much room there is on the cards. If it weren't for the knobs and switches being mounted from the top, this would be really straightfoward.  Here's the caps across the voltage rails on the op amp chips. Hope i did this right.   Channel 12 done. You can see the R37 that was changed (right hand side, just above and to the left of the far right, top cap).  Channel 12 back in, nice new fat caps.  Puny old psu cap  New hotness.  One down, 11 and the control room card to go. The card took me about 45 minutes to recap. |
|
|
|
Post by matt@IAA on Dec 9, 2017 14:32:59 GMT -6
Ha. Found my first mistake. The R37 resistor is only supposed to be changed on the channel boards 1-8.
|
|
|
|
Post by jimwilliams on Dec 9, 2017 17:08:34 GMT -6
I use LME49740NA's in those. Then it's clear and very quiet too. Most all the el coupling caps can go. Check the DC offsets on the output pins (the corner pins 1,7,8,14), if gone, short them out. If you want it back in snip the jumper.
Those green chicklets EQ caps are changed to Wima FKP-2/MKP-2. The super EQ is very hard to refit.
You also must have a small cap in the opamp feedback loop or it may oscillate. 22 pf across 10 k ohms is good. 47 pf across 5k, 10 pf across 22k and so on.
On those .1 bypass caps I see a ground trace on a couple of the + input pins. Dump those large Wimas and install a pair of small mono ceramic .1 uf caps from those power pins 4 and 11 to that ground point. Clean off the pcb with flux remover and a toothbrush. Then you can see your mistakes.
|
|
|
|
Post by matt@IAA on Dec 9, 2017 22:11:10 GMT -6
Hi Jim, I actually came across several of your posts researching this. I was thinking about swapping out the control room card and the 7/8 U2 (for the 2 track summing) for the LME49740NA but I couldn't find them in stock - apparently TI stopped making them? And the ebay ones are all from China. So I decided to roll with it as is.
You're saying if you switch to the LME49740NA you don't need the coupling caps at all any more? Interesting. I thought you always had to have one after an op amp for DC filtering? I am no expert and it has been many years since my few EE courses in school.
I'm only planning on recapping the electrolytics mainly because I had the same thought as you about the looks of that super EQ board.
It's clear this board has been serviced before and at least some of the caps have been replaced over the years - always same capacitance values, but some have higher voltage ratings than what the original schematics call for. If I understand correctly, increasing the capacitance value should decrease the corner frequency, so increasing them like this should only help.
|
|
|
|
Post by matt@IAA on Dec 9, 2017 22:17:44 GMT -6
Control room card work done, going backwards 12-8 done. Not too bad, but I'm not really looking forward to re-installing the knobs.  |
|
|
|
Post by johneppstein on Dec 10, 2017 22:53:33 GMT -6
The stock chips are really quite obsolete - very low slew rate and high noise by modern standards. Not to mention DC offset. The 741 series was the first generation of audio opamps (IIRC) and are well obsolete at this point.
|
|
|
|
Post by matt@IAA on Dec 11, 2017 15:25:39 GMT -6
The whole board is obsolete. That’s why it’s fun!  |
|
|
|
Post by wiz on Dec 11, 2017 15:52:52 GMT -6
cool project
thanks for sharing...
cheers
Wiz
|
|
|
|
Post by swurveman on Dec 12, 2017 9:05:56 GMT -6
cool project thanks for sharing... cheers Wiz It would be interesting to hear an a/b recording of the same song using the same instruments before/after the changes. |
|
|
|
Post by matt@IAA on Dec 12, 2017 11:30:16 GMT -6
Yeah... I originally planned characterize each channel with RMAA before I started. I had already removed the bus connectors to pull a card to verify the schematic and make sure there were no revisions, and I am really nervous about those connectors breaking with age / brittleness. I could manually wire power to each card and check, but I just got lazy. So I'll characterize each channel afterwards and brag about how good it is.  |
|
|
|
Post by matt@IAA on Dec 12, 2017 21:54:31 GMT -6
Got down to channel 5 done tonight. Also realized that I soldered a cap backwards on 8, so I did 100% check and found one more on 7. Having channel one out in front of me for reference / quick check from now on.
I also swapped all of the XLR to Pin 3 hot down to channel 5, flipping them as I go. It's easy enough to just swap the pins on the molex connector (flip J1 1 and 2).
|
|
|
|
Post by matt@IAA on Dec 14, 2017 9:52:01 GMT -6
Alright then. Ready to put the rails back in for testing.  Waiting on my 48v phantom power supply. I'm gonna test a few different out transformers to see what kind of flavor I can get out of them. |
|
|
|
Post by matt@IAA on Dec 15, 2017 10:23:57 GMT -6
Every channel tested ok. Not much measurable difference from before (I only tested channel 1 before) but all channels are around +.05/-.3 dB multitone and +0/-.2 swept sine, with THD in the .006% range.
I have a loose connection somewhere on the XLR on channel 12 so I need to figure that out.
I did get a weird result on intermodulation distortion on channel 1...it's showing like 1.8% intermodulation + noise vs .009% before. I need to figure out if that was a bad test or if there's something going on there.
Edit: Channel 12 was one of those tiny wires they used on the PCB to input transformer broken. Sucked the old solder out, stripped and reinserted, good as new.
Channel 1 was some kind of testing artifact with RMAA. I redid it and everything is fine. So, shockingly, only three errors on the whole thing and two caught during!
|
|
|
|
Post by johneppstein on Dec 22, 2017 2:00:33 GMT -6
The whole board is obsolete. That’s why it’s fun! You don't understand. By "obsolete" I mean noisy, low slew rate, which limits the high frequency response, and poor headroom. None of these traits are in way desirable for an audio mixer. None of that is "fun". They need those coupling caps because they have a high DC offset - they leak DC. The coupling caps induce phase shift. The 741 series was the first family of opamp chips that were at all useable for audio. They were obsolete way back in the late '70s. Back then they were commonly replaced with TL07x series chips, which were a high slew rate Jfet chip, which were much better but still primitive by today's standards. |
|
|
|
Post by EmRR on Dec 22, 2017 7:51:18 GMT -6
Yeah. The transformers may impart some fun. I haven't heard an obsolete IC that did.
|
|
|
|
Post by matt@IAA on Dec 22, 2017 11:28:16 GMT -6
People have such strong opinions on these boards and it’s always about the ICs. From what I can tell, the Harris 4741 is not the same thing as a 741 family chip. The manual for the 1280b notes "Each I/O board contains 3 IC...4741 quad ICs (not to be confused with the 741 IC)".
For example, PRR on groupdiy says
“If something else sounds way-better, I'd have to wonder. The TL074 has similar noise, marginally better speed, and worse output drive...NatSemi wants you to consider LME49740, which indeed has better numbers; in the EQ you have to wonder if a 55MHz amp connected to that many-pot harness proofed with 3.5MHz amps might find a way to oscillate in the CB band....Don't throw-out the 4741s. ”
Another guy on the purple site says "Though the Harris/Intersil HA4741 seems to be a quad 741, nearly every parameter of the Harris chip is better than those of a regular 741.
I once had a RCA mastering EQ on the bench and was wondering why that thing sounds so good. From reading the chips I thought crap, these are 741s - it´s not possible that it sounds good. But crosschecking the datasheets revealed that these Harris chips aren't too shabby for the time they were coming from. A 2520 doesn't look much better on paper...."
I see that the spec sheet on an LME49740 is night and day, 20V/us vs 1.6 on the 4741. On the other hand, I like the way it sounds now and making it better is subjective, especially considering the cost and effort if I have to add an external psu, feedback caps, and so on (plus the chips themselves). And if you really wanted to upgrade to some of the modern chips I’d need to go with a socket adapter. Silk purse, sow’s ear and all that.
A friend of mine is making a simple circuit similar to the pre and eq of one channel and were gonna do some chip rolling for funsies. I'm going to try a LME49740NA in there and see what difference it makes.
|
|
|
|
Post by matt@IAA on Dec 27, 2017 21:08:46 GMT -6
Soon...  |
|
|
|
Post by rocinante on Dec 30, 2017 16:43:10 GMT -6
Have you considered using Jeff's (capi) ACA summing rig? Modern op amps are going to open that guy up and only make things better. I think swapping the Ic's would have the best effect. Well Jeffs summing bus would probably have the best. API channels also use 47 ohm bus resistors.
|
|
|
|
Post by matt@IAA on Dec 31, 2017 14:51:57 GMT -6
Have you considered using Jeff's (capi) ACA summing rig? Modern op amps are going to open that guy up and only make things better. I think swapping the Ic's would have the best effect. Well Jeffs summing bus would probably have the best. API channels also use 47 ohm bus resistors. The concern I have with changing ICs is oscillation and needing to upgrade the power supply. Thinking of that as a round two later. I’m not sure I have heard of the CAPI aca rig. Can you explain? |
|
|
|
Post by wiz on Dec 31, 2017 17:56:35 GMT -6
Have you considered using Jeff's (capi) ACA summing rig? Modern op amps are going to open that guy up and only make things better. I think swapping the Ic's would have the best effect. Well Jeffs summing bus would probably have the best. API channels also use 47 ohm bus resistors. The concern I have with changing ICs is oscillation and needing to upgrade the power supply. Thinking of that as a round two later. I’m not sure I have heard of the CAPI aca rig. Can you explain? Thats a really good point... having some experience down that road myself... its nowhere near as simple a task as people think.... cheers Wiz |
|
