Opamp Labs - Inside the can

[Ethan]

Definitely a good call!  I've never tested this but perhaps resistance values would still hold up to measurement after the outside is "burned" in acetone? ...I have no idea   But whether their resistance values would be reliable after exposure to extreme heat might also be a question...?  I'm sure someone with more experience will chime in soon...

 

[dogears]

4009 schematic. I have question marks in the schematic for the 425 posted above based on the "corrected"/drawn in values in this schematic. Two values for some parts, parts drawn in, etc.

This is a larger image than the original the guy sent me. What I thought then was an emitter follower after the diff pair is actually a pnp common emitter stage. The arrow is clear in this one.

I'm a little bit confused, trying to follow this.

In the first one, you have a voltage divider set up by the 150/100K from VCC+ to 0V, which is connected to the output of the 2N3494 / BC548.

In the second photo, that same pair of resistors is biasing the base of the 2N2484 differential pair at the +IN, marked pin 8.

At any rate, I can't quite understand exactly what is supposed to happen after the second differential stage.

 

[rackmonkey]

Right. Like I said, what happens inside the 4009 can is by no means clear from the schematic the guy sent me. If you look at the bottom of that image, there’s a drawing that he told me he thought was correct. I don’t think he was very sure though. I intended to simulate the different circuits/values in that image prior to posting the schematic to try to surmise what is likely correct. But if I waited to get around to that, it’d be another month or two before i’d Be able to post anything. So I figured we’d iterate through the possibilities here and hopefully use the group brain to come to a conclusion.

But you’ve zeroed in on the most problematic area, I believe.

 

[rackmonkey]

dogears, I went back and looked again and you’re right that I left off the connection from the 150/100 to the base of the 2N2102.  I’ll fix that. Thanks.

 

[rackmonkey]

Missing connection added.  It's really close to the VCC+ rail, but it's separate from it. Was trying to show that it's actually inside the 4009 (pink outline), but that was probably not the best thing for readability.

 

[dogears]

There is a mistake there somewhere, but I can't find it.

 

[rackmonkey]

Have a look at the 4009 schematic I posted. The “mistake” is there. The discrete front end diff pair and output complimentary pair - everything outside the 4009 opamp - is the stuff I traced out from the PCB. That’s accurate. All questions here relate to what we can’t see inside the 4009 can. Everything we think we know about it is in that schematic.

I’m on the road until tonight, but I’ll mess with it in LTSpice when I get home. Won’t spend a lot of time on it though.

 

[MisterCMRR]

I know this is an old thread, but it triggered memory of work I did while chief engineer at Quad-Eight in the early seventies. A customer brought in a Losmandy 4009 or 425, I can't remember which, and asked if I could explain why it sounded so bad when operating at high level. I tested it and looked at its output on a 'scope. What's peculiar is that, if it was ever driven hard enough to clip - if only for a micro-second (which would never be audible), it stretched the +rail output voltage level for almost 100 micro-seconds (which is very clearly audible!). Losmandy apparently didn't understand transistor saturation. A look at your schematics for both the 4009 and 425 shows no emitter-to-base resistor on the PNP stage (gain stage/level shifter). Once that PNP saturates, there's nothing to pull current out of the B-E junction, so it just "sticks" on until the charge bleeds out. Good engineering practice would dictate putting a resistor between B and E to consume about 30 to 50% of the collector current that's driving it. It appears his input pairs operated at about 100 µA each, so an appropriate resistor would be about 0.6 V / 30 µA = 20 kΩ. At the time I did the testing, I didn't know the schematic but guessed it was something like this. Drives home the importance of watching the way amplifiers overload - they should do so gracefully, doing nothing more than flattening the top and bottom of a sine wave. Try it on your 4009 and 425 amps ... drive a 10 kHz sine wave slightly into clipping ... you'll see the top of the sine wave "stick" to the + rail for almost a full cycle!

 

[JohnRoberts]

In power amp design it was good practice to use anti-saturation diodes to prevent bipolar drivers and bipolar output devices from saturating. It takes time and negative drive to discharge base-emitter junctions.

JR

 

[rackmonkey]

Great insights as usual, Bill. Since the time just after this tear down when i built a few channels of 425-based preamps, i have come to the conclusion that they‘re useful and sound great - until you push them hard! Now i have an explanation. Something there IS audible! I‘ll try to find time to do a capture at 10 kHz to see the sticking you’re talking about.

At the time I wrote this up, I didn’t have enough experience recording with them to have noted that. But you can definitely hear it. I stand by my assertion that these are easy to build and useful “color” preamps, but you can‘t expect to get usable, “more color” results by nudging them just over the edge in the way we typically do. Just my practical experience. And now you post this, which is interesting, even though you state that the effect shouldn’t be audible. All i know is, something is happening there that gets in the way of expected operation at the edge.

Your analysis motivates me to finish an analysis of the 360BM. From what i can see on the PCB so far, similar architecture, but (tiny) transformer-coupled input and single-ended output stage. I always thought those sounded a little too wooly and veiled for my taste.

Superficially and on quick glance, the 425 has a similar architecture to your AM-10. But the small details matter. I’d love to hear your thoughts on some of those classic Quad Eight designs one of these days - what drove certain design decisions, what compromises had to be made, and what you would do differently (if anything) with the same components at your disposal now. (Other than incorporating the breakthrough approach of the Outsmarts architecture of the 1646, which is obvious!). It would be educational, I’m certain.

thanks again!