do apex audio amp ic's have some kinda special rules?

[coppercone2]

I have 5822 and 5819 in large box 2

vdrop is 0.178 on the 5819

maybe I should wait till 4am to buy another apex so i can make a comfortable bad late night decision

soldered them in under a nearby ceramic decoupling capacitor close to the shark cage. anything can happen now

[coppercone2]

alright, I put the diodes on, got the part out of the trash and fixed the leads with a pair of non esd safe knipex pliers.

it looks like in circuit now the input and the output are both at 0V and current consumption is low. I need to plug in signals still but either

1)the diode eliminates a startup problem
2)i broke it more
3)i fixed it by handling it like a troll, did I mention the diode polarity was wrong on first attempt, and the negative voltage rail got positive voltage because I swapped the PSU leads to red black red black rather then red black black red. I assume the zener diode did some work for the first time in its life

but its not stuck at 14V on the output.

none the less a 0 signal is less infuriating then a next to rail signal even if broken, so it broke in a better way

its possible the PSU was causing the same problem as it was in circuit with 2x regulators (since that is what it is) during benchtop testing with leads

its kind of ridiclous though about the diodes. they need to add that info in LDO sheets and stuff. like it must be damaging the market not to know if you need low drop diodes there for no good reason. before it requires a 2000 dollar FPGA with polyphasic dsp quantum bullshit to solve a problem because of missing diodes

[coppercone2]

i reattached the 50 ohm termination resistors on the input. no current draw

I think those magic diodes did something

[coppercone2]

hmm I don;'t think tis outputting current.


but not having 14v on the input seems promising. i need to wire a load test. maybe i will get another one. i just need to believe those diodes will prevent a power on death

the circuit i have is a lttle weird but with the pa60 it was smoking... power resistors with some seriously ghetto wiring

[coppercone2]

Ok, so what I am seeing now, is that when its set as a inverter, only whatever bleeds through from the function generator is making to a load.

I disconnected the feedback loop (resistor from turret) and jumpered the output to the inverting terminal to make a in circuit buffer amp. It does not appear to do anything.

I wonder if those diodes helped or if it just totally broke down inside.

[jonpaul]

IC body substrate diode latch up is back 50 years.

All bipolar ICs have it. Larger hybrids like the Apex power op amps are more suseptable due to large area of the transistors.



Dual supply opamps of  any type need a reasonable supply turn on.

Check the turn on of the setup, +/- PS.

Try with a good quality lab supply.


Bon chance

Jon

[coppercone2]

like i said, its regulated locally (island regulator principle)

i used the same damn method for many circuits with no problem. I think its just the power op amp that needs special attention.

is there any other things i can add to a op amp PCB to make it less prone to break on startup? I got the diodes now. When I could not figure out WTF was going on, I cut off the ESD diodes because I forgot the resistors under the PCB were there, I thought the diodes were getting hot. I will reattach those too.

To summarize for anyone else trying this: the effect of not having the SK diodes on the rails might be that the op amp inputs are capable of sourcing rail current through something or another.


the power supply is LQD422. I bet its the local regulators and not the supply. I think that one is really symetrical. their identical sections, powered from a split transformer.

I am also gonna make some SK diodes on lugs so I can put it in front of the power supply for testing with cables.

[coppercone2]

Lol ok

On the PCB,
-both channels don't seem to put out current, not high power consumption. the output is 0V. Basically does not respond to input on either channel, you get some feedthrough of the signal (same as powered off) on the output.

on wires with the SK diodes at supply.
-1 amp on both chips works. the other amp is at rail voltage, Does not respond at all to any activity on the non inverting pin. . they can smoke a resistor. if you connect a resistor to the railstuck output pin, it gets serious juice through it. The other side of the amp gives output proportional to input.


IDK if the supply is causing a channel to seize. The board has lots of decoupling (good capacitors) and diodes right near the chips.
But it seems if I do the same experiment where i short out the feedback path to make a inverting amp, its not putting out current. 

so IDK if anything is even broken. Maybe?

when I try it from the PSU I have the 2 SK diodes on each channel and then a long tight twisted pair of 18 AWG wire connected to graBBERS. Maybe I need local decoupling or something. I think I need to buy another socket, and make a test PCB with some decoupling capacitors.   

christ, this is about as easy as starting a old chain saw


Maybe I get the old HP dual supply with the 'tracking' function out. Then raise the voltage slowly for the chip.

bcause the version of this circuit that worked was
1) older amp
2) protoboard
3) decoupled pretty good (tantalum and ceramic). The PCB version is ceramic and electrolytic polymer (did not want tantalum).
4) same power supply

Maybe it does not like ground planes or something but still wants decoupling?? or tantalum caps? maybe I made the rail decoupling ESR too low, I did go kinda crazy with it....

[coppercone2]

lol now with the dual HP supply that has the tracker it looks like the OLD apex chip is stuck with a rail output but the newer chip is pretty much doing the same thing that it was doing with the lambda dual supply turned on to set voltage. (in PCB)

they keep changing states!


AND IN THE circuit with the new supply I get a symetrical response but its not correct. It seems to have trouble pulling negative. I wonder where the damage occured if the chip was damaged by the mechanism proposed. maybe it causing that. and there is at least a different response when the power is applied and removed. before it was like invariant.

I am actually thinking way back when I did have filter circuits that did not work. Not SC filter but still. I wonder if this issue was related. I think I have some research to perform on circuits abandoned long ago. They got so complicated with dead bug soldering I never wanted to take it apart to verify the op amp did not break on start up, I just gave up.

[coppercone2]

also, what is some good integrity test to run on power op amp to check for damage?

like most informative tests, etc.

I think I gotta take this slower so I am on a cloaked goose hunt

I thought just inverting/non inverting amp with a resistive load. But are there more telling tests? perhaps just a speaker maybe? good quality audio = good op amp?

[coppercone2]

well the high speed scope shows that something funny is happening when you have a high bandwidth.

a test you might want to run on bad amplifiers is to sweep the frequency low pass filter (i.e. rigol) to see if the wave form changes or gets less stable

[coppercone2]

bought another socket so i can try these on a slowly ramped tracking power supply with diodes in simple configuration

[coppercone2]

Ok I got another socket and I made inverting amplifiers on both sides. It looks like the same part of the opamp is fried on both. I soldered resistors (grounded on input) to the socket to see if 0V - 0V. One side works one side no. It makes me think my board has some kind of defect thats waxing parts

for the non inverting topology, one side responds to input being grounded, the other does not.

I wanted to test it by
1) connecting it to 0V output dual tracking HP power supply
2) slowly ramping it via knob while in tracking mode to +-10V
3) dismantle socket configuration and rewire it to non inverting topology and repeat 1&2


Can I do this without the damn diodes and capacitors to just see if a new in box chip works with grounded input? I verified the supply tracks with 1mV accuracy

Wires soldered to socket that are inserted into binding post. the only thing to clip on is oscope lead. connection lenght 6 inches of wire per supply connection (3 wire)

the idea is to get the part from the box while wearing a esd strap, insert it to socket, ramp up supply. When powered and set to 0V the supply measures 4.1 milivolts. Though I did not see any overshoot at all during turn on so I guess I can plug it in off too.

[coppercone2]

I made a dedicated test socket for inverting amp configuration. This one I plugged in when it was set to approx 2mV and ramped up to 12V with potentiometer.

The third apex chip functions as expected. 0V in 0V out. Hard soldered. The other two are broken.


I will rewire the socket just to test non inverting operation, but I suspect either my PCB is breaking parts, or I got two broken parts in a row.

Given the fact that both amps have fried at the same position, I am inclined to blame my PCB. Not sure how its doing it.

Hahaha and I thought that this board was simple enough to test! Tempted to put the diodes locally right up against the IC and test it again but its expensive proposition.


I can wire in a dummy load, and compare to the broken chips in the PCB, to check for latchup conditions. I should have tested the parts first before plugging them in before going on this wild goose chase but I was so confident it would just work.

seems to work in non inverting configuration too.