M670 compressor

[duantro]

It's probably going into thermal shutdown.  How much voltage is across it input pin to output pin ?  The more voltage that is across it the hotter it gets and the less current it can deliver.  Looks at the data sheet there is a graph for this phenomenon.

Thanks Rob,
I'm getting 5vdc on input pin, output pin is grounded per circuit. The data shows up to 10vdc differential from input to output, if I'm reading it correctly. I'm using an Antek 2215 for the heater circuit. https://www.antekinc.com/content/AS-2215.pdf

 

[Kingston]

duantro,

you have to distribute that heat somewhere away from the regulator. Sure that chip claims 5A regulation, but that is only in perfectly optimal heat-sinked conditions. Ideally that lm338k should be left to drop exactly 3VDC. That ensures minimum heat across the device. 10VDC differential is simply too much for the amps you are drawing. A good way to deal with this is a big-watt wire-wound resistor to drop a good 5-7VDC _before_ it enters the regulator. You have to calculate the exact resistance and required resistor watts based on expected heater current draw. Often it may turn out you never really needed the regulator at all. A perfectly matched big-ass wire-wound resistor (or two in series) can sometimes produce stable enough VDC without noise.

This is in no way specific to M670 but applies to all tube DC powered heaters since it's usually impossible to get a perfectly matching transformer for the job.

 

[duantro]

duantro,

you have to distribute that heat somewhere away from the regulator. Sure that chip claims 5A regulation, but that is only in perfectly optimal heat-sinked conditions. Ideally that lm338k should be left to drop exactly 3VDC. That ensures minimum heat across the device. 10VDC differential is simply too much for the amps you are drawing. A good way to deal with this is a big-watt wire-wound resistor to drop a good 5-7VDC _before_ it enters the regulator. You have to calculate the exact resistance and required resistor watts based on expected heater current draw. Often it may turn out you never really needed the regulator at all. A perfectly matched big-ass wire-wound resistor (or two in series) can sometimes produce stable enough VDC without noise.

This is in no way specific to M670 but applies to all tube DC powered heaters since it's usually impossible to get a perfectly matching transformer for the job.

Thanks for this info! I'm going to yank the LM338T and put in a LM338K like recommended, I just ordered the wrong part. I see that the LM338T is only rated at 25wattts. I'll explore a wirewound resistor in front of the lm338K also.

 

[Holger]

I've recently finished my version. 4 RU glossy dark blue painted front panel, external PSU. Still needs some tweaks but I like what I hear so far.

 

[Bernardo]

hello everyone, this thread is quite old but i have started to build this compressor and i have already found a pretty basic problem with the LM338K. after reading the datashet of the regulator, is it possible that the pins are changed in the schematic? input would be 2 and output 3? I have the problem that I have to put the 1N4007 that goes from 2 to 3 in the opposite direction to have -12.6v at the output of the regulator, but when you insert the tubes, it burns. Maybe I need 2 of these in parallel? Thank you.

 

[scott2000]

is it possible that the pins are changed in the schematic?

I'd seriously be surprised if that were the case. Speaking of case, how is it mounted?

 

[Bernardo]

is this correct?

 

[scott2000]

Pretty certain you are right about the schematic pinout numbers being wrong for the lm338k . It's inputs are pin 2 and the to-220 lm338 inputs are pin 3.

 

[Bernardo]

What is the feedback/feedforward option for? And 25k x2 potentiometer,what do it?I don't notice compression and the meter moves very slightly, unit sounds great
Thanks

 

[scott2000]

What is the feedback/feedforward option for? And 25k x2 potentiometer,what do it?I don't notice compression and the meter moves very slightly, unit sounds great
Thanks

Someone may correct me but it looks like the feedback and feedforward is an option to select where the sidechain signal comes from. Either right at the input, or right at the output. You can look up feedforward/feedback compression and it'll explain.
The 25k pots look like maybe a threshold but may be called something else. For how much signal you want to go to the sidechain/control amp. Info probably somewhere. Maybe the pm660 thread has some relevant clues even.

 

[Bernardo]

Thanks a lot scott.

 

[Bernardo]

Wow, what a headache, I've been thinking about the same thing for a month. The compressor sounds perfect, but it only compresses 2-3dbs. I've checked everything like 20 times and it's well connected. my current values are 270v for the input to the 5687 and 180v for the inputs to the 6N5P. -6v for the 1M input of the diode bridge, and -3v for the output of the D45H11. the audio transformers all measure the same. the primaries of the 600:10K. They measure 26R and the secondaries 450R. 10k:600R transformers measure 948R primaries and 120R+120R secondaries. Any clue please, I'm a simple fan who is lost. thank you

 

[Script]

What happens if you feed a louder signal into the unit? Does it compress more then?

 

[Bernardo]

no, it saturates the input but the compression is just as light.

 

[scott2000]

no, it saturates the input but the compression is just as light.

Guess there maybe needs to be more detailed info provided. Looks like Kingston said this has high headroom and low distortion in the beginning of the thread?

 

[Bernardo]

my problem is that it only compresses 2 db with the 22k potentiometer at full

 

[scott2000]

my problem is that it only compresses 2 db with the 22k potentiometer at full

Post #102 mentions some info that may be relevant? IDK...

 

[Bernardo]

I'm going to post how I solved my problem in case one day someone has the same headache. The unit was built according to Mr. Kinston's schematic except for the LM338 regulator. After thinking about the position of components and values over and over again, changing tubes and transformers, the solution to my problem was to lower the voltage of the heaters from 12.6v to 6.3v. From there everything started to work apparently normally. I don't know if it sounded similar to a fairchield, because I've only seen it in photos. greetings to all!

 

[duantro]

I recently have been getting some clipping distortion that is mainly noticeable from the low end frequencies, ONLY WHEN I have the 600r T pad up past 11 O'clock with loud signals. I pulled the tube boards to double check that I swapped the input grid resistors at PL1 to 47k. I had done that, but noticed the Plate resistors RSA and RSB were burnt to a crisp on both left and right boards! I swapped them out with 1 watt 4k7. I'm running that Plate supply at 200vdc. Unit works and sounds good, but still is sensitive above 11 O'clock. These don't need to be 47k also do they?
Perhaps the cooked plate resistors damaged the tubes. I’ll swap them out and see if it helps.
Edit: I changed the tubes and it’s still doing it.