Symmetrical FET compressor.

[Gustav]

Small update, since I got some missing parts yesterday.

Measured my AC supply today (40 across, 20V to each tap from center)

Connected it to the PSU without the 18+/18- jumpers connected to the circuit from the PSU. Got 18+/18- DC coming out.

Made the jumps (no smoke!). Output transformers mounted, ICs omitted. 18+ looks fine, 18- reads 9-.

Will cut the PSU and try it loaded to something I know is working, and if it is, I will check if I simply need to mount the ICs to get the -18 rail going. I had a similar problem with an earlier project, where the voltage was loaded down on diodes going to the ground rail/the output trafo mounted.

Not really in a rush - still trying to find a NOS dual FET, so I can shoot it out against candidates for replacement.

Does anyone know this shop? https://www.studioelectronics.biz/sunshop/index.php?l=product_detail&p=1176

Gustav

 

[Dr Gris]

Does anyone know this shop? https://www.studioelectronics.biz/sunshop/index.php?l=product_detail&p=1176

Gustav

That's David Kulka's place. Regular poster here. Serious stuff. Contact him.

//M

 

[Gustav]

Does anyone know this shop? https://www.studioelectronics.biz/sunshop/index.php?l=product_detail&p=1176

Gustav

That's David Kulka's place. Regular poster here. Serious stuff. Contact him.

//M

Ah, I feel like an idiot for even asking now (It even says David Kulka just under the logo and in the browser bar)

Will order a few

Gustav

 

[gyraf]

Output transformers mounted, ICs omitted. 18+ looks fine, 18- reads 9-.

Try with output driver IC mounted.

Then look (feel) for something getting hot.

Check for wrong-orientation electrolytics in negative supply line

Jakob E.

 

[ijzerpro]

reverce C25

 

[Gustav]

Thanks so much for the input on C25. (I attached the schematic, which is a bit of a mess, since I have all the pins for the switch-PCB in there)

I also found R19, R20 and R22, which should be 30, not 30K, and C22, which should be 1 uF

Status:

- I now have a stable -18 supply, and a positive supply that fluctuates between +17 and +18 with the GR switch out.
- Both the positive and negative supply move up and down about 1 volt with the GR switch in.


Still lots to figure out, maybe even a misinterpretation of the schematic - Calling it quits for today and putting in some real time tomorrow

Gustav

 

[Harpo]

From your schematic at least C25 and C32 are still the wrong way round. Jakob already hinted electrolytics (plural).
Your PSU section reference voltage uses a different symbol and name [AGND] from the rest of the circuit [GND].
For usual the AGND symbol is used for safety ground/chassis ground and XLR-pin1/shield. These different AGND and GND potentials might join at one single spot (Urei did it at the input terminal).

 

[Gustav]

From your schematic at least C25 and C32 are still the wrong way round. Jakob already hinted electrolytics (plural).
Your PSU section reference voltage uses a different symbol and name [AGND] from the rest of the circuit [GND].
For usual the AGND symbol is used for safety ground/chassis ground and XLR-pin1/shield. These different AGND and GND potentials might join at one single spot (Urei did it at the input terminal).

Thanks!

The board I am working on is based on the current schematic, so I listed the errors without changing the schematic. I only listed the errors caught on one side as some sort of habit from looking at the original schematic. they have been fixed on both sides. Should I change the schematic even without a new board revision? 

The AGND symbol is there due to an autoroute-experiment I did putting a resistor between two different GND nets. I ended up manually routing a star ground and simply connecting the two ground symbols directly in the schematic (further along in the schematic). I'll fix this later, when I clean up the schematic.

I just spent a few hours reading about diodes on my balcony in the nice winter sun. The original schematic calls for 10v Zeners, and I randomly swapped them to 1n4148 because...well...just because. Will dig out some zeners.

Thanks again for taking a look!

Gustav

 

[Harpo]

The board I am working on is based on the current schematic, so I listed the errors without changing the schematic. I only listed the errors caught on one side as some sort of habit from looking at the original schematic. they have been fixed on both sides. Should I change the schematic even without a new board revision?

I was referencing your latest schematic link. In Eagle both board and schematic reflect the same. When you fix errors in schematic, double check your already routed board layout. At least the silkscreen will change.

and simply connecting the two ground symbols directly in the schematic

Doing so, you already got a warning from Eagle for joining these nets.

The original schematic calls for 10v Zeners, and I randomly swapped them to 1n4148 because...

Only in the rectifier section in front of the attack/release circuit.
The diodes in the missing metering-amp and GR tracking/meter zero section (above the psu from original schematic) can be 1N4148s IMHO.

I'd miss the power-on LED (D80 + DS5 from original schematic).
Two 1N400x diodes between +18V/GND and GND/-18V (you probably calculate the voltage setting network a little below 18V to not exceed the abs.max. rating of TL07x) might prevent from voltage regulators hanging on power up, a common GSSL fault. You could use these -for +/-18V not needed- D19 and D20 for this purpose. Adding maybe 10uF/35V caps in parallel to R92 and R93 should help for a lower noise supply.

 

[Gustav]

The board I am working on is based on the current schematic, so I listed the errors without changing the schematic. I only listed the errors caught on one side as some sort of habit from looking at the original schematic. they have been fixed on both sides. Should I change the schematic even without a new board revision?

I was referencing your latest schematic link. In Eagle both board and schematic reflect the same. When you fix errors in schematic, double check your already routed board layout. At least the silkscreen will change.

What I mean is, I made the changes on the physical board and wrote the notes (in the thread as well).

I am reluctant to change the schematic/make changes in the board layout as long as I am working on this prototype, and unless there is some standard for doing so, I think the errata notes along with the schematic being corrected is the most logical way to go - then change the schematic and generate a new board which reflects those changes.

But I do not know what common practice is...

and simply connecting the two ground symbols directly in the schematic

Doing so, you already got a warning from Eagle for joining these nets.

For the original experiment, the AGND could have been named "BLUPR". It will be cleared up in a final schematic using a common symbol, but I only see a semantic problem as is!? Since you mention it again, I fear I am failing to understand some fundamental error.

The original schematic calls for 10v Zeners, and I randomly swapped them to 1n4148 because...

Only in the rectifier section in front of the attack/release circuit.
The diodes in the missing metering-amp and GR tracking/meter zero section (above the psu from original schematic) can be 1N4148s IMHO.

I posted the reasons for omitting the meter buffering circuit and peak/value switch in an earlier post, and again, I fear I am overseeing something fundamental, since I mentioned changing the current 1n4148 to zeners in the previous post.

Two 1N400x diodes between +18V/GND and GND/-18V (you probably calculate the voltage setting network a little below 18V to not exceed the abs.max. rating of TL07x) might prevent from voltage regulators hanging on power up, a common GSSL fault. You could use these -for +/-18V not needed- D19 and D20 for this purpose. Adding maybe 10uF/35V caps in parallel to R92 and R93 should help for a lower noise supply.

Thanks for the improvements to the PSU.

I refer to 18v+/-, but I programmed the regulators to run just under 18V due to considerations on the 074. How low would you go?

Any obvious (cheap, DIY friendly) replacements to the TL074 I could consider for a possible next revision? 

Gustav

 

[gyraf]

Any obvious (cheap, DIY friendly) replacements to the TL074 I could consider for a possible next revision? 

Maybe MC33079?

 

[sircletus]

I've thought about doing something similar for awhile.  I'd love an 1178!  The only thing stopping me (aside from time and money) is the ratio switch.  I'd like to keep the same pushbutton configuration as the 1176, but finding a 4PDT interlocking switch set should prove near impossible.  Glad to see you doing this with a rotary switch!

 

[Harpo]

Gustav, is the backwards connected LL5402 in 1+1:2+2 stepup config instead of primary windings in parallel and secondary windings in series for 2:1+1 on purpose ?
Instead of an additional opamp (IC4), you could use the left over IC2d for the tracking amp as an obvious (cheap, DIY friendly) replacement. No need to change the TL074 for a different part in this circuit IMHO.

 

[Gustav]

Thanks for all the help so far.

The Lundahls are set to step up on purpose, since it worked on the 1176, but as per recommendation from Jakob, I will add a cap going to it. It was also on my short list for things to try out.

Progress since last update:

- I replaced the 1n4148 diodes for zeners.

- I get about +17.5V and -18.5V on the supply, but when I make the contact on the bypass switch, they are both dragged about 1 volt down, and the positive supply slowly fluctuates almost a full volt.

Double checked all pinouts for transistors, and I found no errors.
Double checked all pinouts for opamps, and I found no errors.
Double and triple checked my switch connections, and I found no errors.
Compared mine and the original schematic another time. I found no new errors.

So now I am trying to figure out whats bringing my supply to its knees.

- Bad component?
- Error in the copy of the schematic?
- Error in Urei's schematic!?

I have gone through the schematic so many times, and I have had the help of so many extra eyes, I think it would be best to look elsewhere.

I was going to cut out the metering circuit and start cutting supply traces methodically to the various parts of the circuit today, but first, I tried simply removing the 074.

With the 074 removed, the supply is stable, and, of course, I am getting +18V on both sides of the switch wether the connection is made or broken.

If Im really lucky, its as simple as a bum 074, but today is saturday, so I will investigate further on monday.

The transistor (T1) should still be loading the 18V supply, even if its not seeing anything coming from IC2C on the base, right? What I am basically asking is - can I rule out T1 and its twin as suspects for dragging down my supply?

Gustav

 

[Gustav]

Finding some time to work a little on this one again.

I did another meticulous check of the switches, and yet again, I could find no errors.

I measured DC on the input of the output transformers and got around 0.5V on both channels. To be completely honest, I have no idea how much is expected or acceptable, but I cut the trace and put a blocking cap in there. As a result, the supply no longer fluctuates when I flip the GR switch. 

It still bogs the supply down about 0.5V on each side when I switch in GR, so it is still not healthy, but the voltage is steady in either state.

So, I usually don't bother checking the signal path before I have the supply voltages running, but Ill make an exception and see if it speaks - even if it turns out to be some weird, foreign language.

Gustav

 

[Gustav]

It passes audio fine, responds perfectly fine to input pot adjustments.

It responds to adjustments on the output pot just fine, expect - when I crank it up, it goes into a cycle of dying and coming back again. It dies with a stuttering cough (oscillation?), not a fade out.

Switching GR in and out makes no audible difference (even with the output cranked), but I still measure a voltage drop when engaging it.

I almost had some NOS FETs in my hand, but danish customs made a mistake and sent them back. I think Ill postpone further investigation until I have them.

Gustav

 

[Dr Gris]

Following this with interest. Sorry about the FET's.

 

[gyraf]

motorboating (low frequency oscillation) - check your power supply lines while malfunctioning.

 

[L´Andratté]

Gustav, your learning experiences are very resourceful to me, all this little problems and considerations needed, designing circuits and pcbs. It´s always hard for me to accept that one has to really take his(her?wishful thinking...) time, to get quality.
Just want to say: nice stuff. Might even build it, although atm I´m deeply obsessed with cloning Buchla modules :

 

[Gustav]

Jakob lend me a hand this weekend.

I cut the traces to the output transformers and added a blocking cap at an earlier stage. In transit, one of them had come off, and I had soldered it back to the wrong pin on the output trafo - so, that was an error created by my carelessness, and not an error that was present when the unit was still on my bench.

When that was sorted, Jakob shorted the 100R in the CRC filter, and everything was fine.

At first I thought I had randomly copied the 100R value from a circuit, but in my notes, I calculated the value based on a 50ma load on the regulators (0,05 x 100 = 5V drop), which seemed fine with an 18VAC supply voltage (usually a little more) rectified (about 25VDC) and destined to end up as 18VDC. Obviously, I was wrong.

I appreciate all the help and input received, and the unit is working now.

Only thing left is finding a replacement for the FET. I used an original for the current prototype, but I only have a dozen more.

Gustav