You meant MJL21193/4 ?
Okay, let me guess, you used mjl21193 for mj15003 and mjl21194 for mj15004 ?
A bad trap with these devices, mjl21193 is pnp and mjl21194 the npn !
If you stumbled into this trap, swap quickly these devices !
If this is not the problem it get's likely that these outputdevices are defect...
Have you verified that you have original Onsemi devices ?
Typically these are nearly "unkaputtbar"...
Did you substitute any other transistors ?
Mike
Okay, let me guess, you used mjl21193 for mj15003 and mjl21194 for mj15004 ?
A bad trap with these devices, mjl21193 is pnp and mjl21194 the npn !
If you stumbled into this trap, swap quickly these devices !
If this is not the problem it get's likely that these outputdevices are defect...
Have you verified that you have original Onsemi devices ?
Typically these are nearly "unkaputtbar"...
Did you substitute any other transistors ?
Mike
I found the mistake - I swaped the PNP and NPN type (no it wasn't like that - a few days ago I just put the on heatsink without checking the type just to take the picture of them like that and I just put them on heatsink and forget about polarity 
stupid )...
I fix that and here are new voltages:
Ubb of drivers: 2.07V
Ubb of output tranzistors: 1.4V
Ube of drivers: 0.39V
Ube of output tr: 0.66V
stupid )...I fix that and here are new voltages:
Ubb of drivers: 2.07V
Ubb of output tranzistors: 1.4V
Ube of drivers: 0.39V
Ube of output tr: 0.66V
more measurments:
Uce (BC556 on heatsink)= 2.08V
Voltage on emiter resistors: 81mV (trimer is turned to minimun)
Voltage on saftey resistors: 13.3V
Yep, tranzistors are MJL21196/94!
Update:
when I shorted input with GND Ure droped to ~41mV, and Ube of drivers increased to 0.43V
Output voltage is 4mV
Uce (BC556 on heatsink)= 2.08V
Voltage on emiter resistors: 81mV (trimer is turned to minimun)
Voltage on saftey resistors: 13.3V
Yep, tranzistors are MJL21196/94!
Update:
when I shorted input with GND Ure droped to ~41mV, and Ube of drivers increased to 0.43V
Output voltage is 4mV
Mlaen, you meant MJL21194/3 or MJL21196/5 ?
And yes, do like Jan said, check immediately if amp is functional,
after you have verified that DC-offset is at least below 100mv,
connect a speaker and a source. You can keep safetyresistors for
this test, just check if signal is reproduced !
BTW, you might consider getting another multimeter, it's too easily
fooled by frequencies in DC-signals...
Mike
And yes, do like Jan said, check immediately if amp is functional,
after you have verified that DC-offset is at least below 100mv,
connect a speaker and a source. You can keep safetyresistors for
this test, just check if signal is reproduced !
BTW, you might consider getting another multimeter, it's too easily
fooled by frequencies in DC-signals...
Mike
Yep, that's why i encouraged you with this topology !
Low DC-offset at once, and the strongest/tightest bass i ever
experienced, beating all other topologies i know / tried.
And some more qualities, you will find them !
Okay, you said 81mv over the 0.22 ohms Re's, over both or single ?
You can replace the 500ohm pot with a 1k pot, giving more space
for adjusting, or replace the 4.7k with a smaller one like 4k.
or you can increase the 1.8k to 2k or 2.2k.
As you have substituted all transistors in outputstage, the initial
values from the design might no longer fit.
In my amp, i use 2k instead of 4.7k, use a 1kpot, and 500ohms
instead of 1.8k. But this does not necessarily fit in this one, as vas
current might be different.
Mike
Low DC-offset at once, and the strongest/tightest bass i ever
experienced, beating all other topologies i know / tried.
And some more qualities, you will find them !
Okay, you said 81mv over the 0.22 ohms Re's, over both or single ?
You can replace the 500ohm pot with a 1k pot, giving more space
for adjusting, or replace the 4.7k with a smaller one like 4k.
or you can increase the 1.8k to 2k or 2.2k.
As you have substituted all transistors in outputstage, the initial
values from the design might no longer fit.
In my amp, i use 2k instead of 4.7k, use a 1kpot, and 500ohms
instead of 1.8k. But this does not necessarily fit in this one, as vas
current might be different.
Mike
Oh, i did miss the update in one of your postings where you say
that grounding the input reduces current.
This is a indication for oscillation...
In this case you can stop modifying vbe-multiplier.
You can again install the 100nf in paralell to the 220ohms, and
check if it reduces bias. If it does, your amp oscillates.
Your 10ohms in series with 100nf at output, does it look like the
other 120ohms ? If this one is wirewound, you get real problems...
For testing purposes try a normal 10ohm.
At input, you should have a 100pf - 330pf cap paralell to the 10k.
Also an inputcap is missing in the schematic, but it looks like you
have it on your pcb.
But now it goes to the design itself... It looks a bit undercompensated to me...
The original design might have got away with that because of slower outputdevices,
the next steps get hard, remotedesigning is difficult...
Mike
that grounding the input reduces current.
This is a indication for oscillation...
In this case you can stop modifying vbe-multiplier.
You can again install the 100nf in paralell to the 220ohms, and
check if it reduces bias. If it does, your amp oscillates.
Your 10ohms in series with 100nf at output, does it look like the
other 120ohms ? If this one is wirewound, you get real problems...
For testing purposes try a normal 10ohm.
At input, you should have a 100pf - 330pf cap paralell to the 10k.
Also an inputcap is missing in the schematic, but it looks like you
have it on your pcb.
But now it goes to the design itself... It looks a bit undercompensated to me...
The original design might have got away with that because of slower outputdevices,
the next steps get hard, remotedesigning is difficult...
Mike
Changing just the resistor (10R in zobel) changes nothing, so he's OK.
When I add a cap paralel to 220R Ure=0V (same as if I ground the input). Then I started turning trimmer and in one point, sudentlly I have ~80mV again...
About the schematics...
As you can see from PCB I have 1uF cap in series after 470R res. So I have only CR network, RC is missing...could this be causing oscilations?
When I add a cap paralel to 220R Ure=0V (same as if I ground the input). Then I started turning trimmer and in one point, sudentlly I have ~80mV again...
About the schematics...
As you can see from PCB I have 1uF cap in series after 470R res. So I have only CR network, RC is missing...could this be causing oscilations?
Hmm, i looks to me very nuts to use an inductive resistor in a RC that is intended to damp HF...
How should it do that if the resistors blocks HF ?
And i have read too many times that some designs start to oscillate
if built with wirewound emitterresistors !
And you try to keep wires and leads short, just to use a highly inductive resistor ?
Mike
How should it do that if the resistors blocks HF ?
And i have read too many times that some designs start to oscillate
if built with wirewound emitterresistors !
And you try to keep wires and leads short, just to use a highly inductive resistor ?
Mike
Hi mlaen, the fact that adding this cap in paralell to the 220ohms
reduces current to zero, and that "bias" jumps to a fixed value,
shows that the amp is very likely to oscillate.
The "missing" cap at input should not cause oscillation, but adding it definitely helps.
The problem now is, that this is the point where designing an amp
really starts. Without scope it's not even sure that the amp
oscillates. For the beginning i suggest adding this cap at input,
the ~220pf in paralell to the 10k, and also adding a RC in diffamp.
This RC would sit between the 2 1.8k, and consist of a cap and resistor
in series. I now can only guess, maybe try 1nf+330ohm, or 330pf+150ohm, or else...
But published designs should have enough stabilityreserves to not
start oscillate when some transistors are exchanged. You even do
not use highspeed output devices....
Do you have contact to the designer ? This is now his responsibility !
Mike
reduces current to zero, and that "bias" jumps to a fixed value,
shows that the amp is very likely to oscillate.
The "missing" cap at input should not cause oscillation, but adding it definitely helps.
The problem now is, that this is the point where designing an amp
really starts. Without scope it's not even sure that the amp
oscillates. For the beginning i suggest adding this cap at input,
the ~220pf in paralell to the 10k, and also adding a RC in diffamp.
This RC would sit between the 2 1.8k, and consist of a cap and resistor
in series. I now can only guess, maybe try 1nf+330ohm, or 330pf+150ohm, or else...
But published designs should have enough stabilityreserves to not
start oscillate when some transistors are exchanged. You even do
not use highspeed output devices....
Do you have contact to the designer ? This is now his responsibility !
Mike
You could try to double the miller cap in the Vas. Then again start with the bias from it's lowest value and see if the "jump" to 82mV is no longer there.
Without a scope, you could make an RF detector to measure the oscillations. Put a diode from the amp output to a cap of say 500pF, other side of cap to ground. Put a bleeder resistor across the cap of a few 100 kohms. Put the DC multimeter across the cap. Anything above a volt is oscillations (if you do this with the input shorted of course).
Jan Didden
Without a scope, you could make an RF detector to measure the oscillations. Put a diode from the amp output to a cap of say 500pF, other side of cap to ground. Put a bleeder resistor across the cap of a few 100 kohms. Put the DC multimeter across the cap. Anything above a volt is oscillations (if you do this with the input shorted of course).
Jan Didden
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.