Marantz 2270 bias won't go below 15mV

[rBuckner]

Hi All

I repaired one amp channel which only had one bad output transistor. I used the OnSemi MJ21193/4 pair and up she came, sounding great. The DC offset adjusted just fine but when I went to adjust the bias it wouldn't go below about 15mV when fully CCW. I checked the other channel which had the original Fairchild outputs and whaddya know, the same thing. I double checked to see if I was on the emitters since I know the service info listed the wrong test points and I was. As the pot is rotated clockwise it takes about a 1/5 of a turn before the bias starts to rise which seems like normal action. Oh, I also adjusted the +35 supply.

So on the channel that was already okay, I went ahead and using the best hfe matched NPN/PNP MJ2119x's, replaced the outputs. The problem still remains on both channels. The distortion measures the same on both channels and is what I expect at 1 watt with the covers off (.14%) and it clips darn near equally on each channel into 8 ohms at 91 watts. At just under clipping the distortion on both channels was the usual about .03%. i did a total re-cap and double checked that the ones in the amp are in correctly.

Any clues as to what I should try to look for? :scratch2:

 

[Jon_Logan]

If it has a "typical" bias circuit;

there's a NPN transistor (bias) where the collector connects to the signal line attached to the base of the positive driver and the emmiter of the bias transistor attaches to the signal line (Base) of the negative driver. This bias transistor can be asked to conduct through heating of itself, adjusting the bias pot, and/or heating of a thermistor. With the set off, you can attach a jumper collector to emmiter of the bias transistor. This will short out any bias voltage spread across the bias transistor. If you do this, the voltage measured across the emmiter resistor(s) (bias current) should be zero. If it is not, there's probably some leakage path, which is a little more complicated to troubleshoot and fix.
If the bias does go down to zero, you can change a resistor that is probably in series with the bias pot. My guess is, the bias pot is in the Base to Emmiter connection of the bias transistor. If it is, then lowering (slightly) the series resistor (with the pot) should reduce the voltage spread across the driver bases, for a given bias pot setting.

You really want to be doing this with a VariAc, and/or dim bulb tester setup. Small changes in the bias circuit can make big changes, output damaging changes, in the bias current.

 

[ecluser]

What is your serial number?

 

[rBuckner]

Thanks for the dissertation Jon, it is very helpful! I'm stuck at work today but I will get on it this evening and report back. I remembered this morning that I have another 2270 in the wings and I'm going to start the evening with taking a look at it's bias adjustment action. Again, thanks!

 

[rBuckner]

What is your serial number?

I'll take a look when I'm home at lunch and report back. Thanks!

 

[rBuckner]

what is your serial number?

40391

 

[ecluser]

By chance your serial number is higher than 3900 ! Unlike the SN 1001~3900, the emitter resistors for the power transistors are at the good place in your amp board.

For your amp, replace R784 and R785 by larger value, 5.6 Ohms should work very well. With this change you will be able to reduce the bias current in the output transistors to near 0mA.

According to the service manual, the bias transistor (H760) and the bias diode (H761) are thermally coupled to the power transistors through the heat sink. It was a bad idea for an amplifier of this topology (complementary darlington). I recommand that you transfer these temperature sensitive biasing components to one of the driver transistors (H758 or H759). Your bias current in the output transistors will be much more stable after this change.

 

[rBuckner]

Thanks ecluser! Those resistors were on my list to check anyway.

I did the experiment Jon suggested, to short out the biasing transistor E to C. That brought the emitter to emitter voltage down to 2.5mV. After removing the jumper I powered the unit back on and the bias was at 2.5mV to start an slowly rose to about 10mV, like over a 30 minute span. So there's still 6mA plus flowing it would seem.

I grabbed the other 2270 I had, popped the covers and ran it 10 minutes or so and touched up the DC offset. The checked the bias on both channels which were both about 10mV. Both went to 2.5mV or so with the pots fully CCW and sat there steady. Both adjusted just fine, with about the first quarter of the pot rotation not having any effect so that ws normal as I suspected.

So something is slightly amiss in unit #1. It will be interesting to compare voltages and resistor values and dink with the resistors that ecluser suggested. I will bump this thread when I get further. It will be a bit as I have a 2285 restoration to start in on for someone.

Thanks for the helpful replies fellas! :thmbsp:

Randy

 

[sregor]

Had a similar problem once with a 4300 receiver (similar output design) (about 25 years ago so YMMV) Couldn't get the bias down in one channel. Apparently Marantz designs the circuit with low tolerance to transistor variation. Turned out to be the VBE drop of the replacement drivers (not the outputs) was about 100 mv less than the original parts. I simply measured the b-e junctions of a couple of different subs, and used the ones that measured highest and everything worked fine. Hope this helps. Steve

 

[Jon_Logan]

Had a similar problem once with a 4300 receiver (similar output design) (about 25 years ago so YMMV) Couldn't get the bias down in one channel. Apparently Marantz designs the circuit with low tolerance to transistor variation. Turned out to be the VBE drop of the replacement drivers (not the outputs) was about 100 mv less than the original parts. I simply measured the b-e junctions of a couple of different subs, and used the ones that measured highest and everything worked fine. Hope this helps. Steve

Increased Vbe drop would (should) definitely help here, without other circuit modifications.

Shorting that driver bases together styill produced a voltage drop across an emmiter resistor(s) though. Something's weird about that.:scratch2:

 

[ecluser]

Marantz 2270 Bias, a case study

After my previous post I was asked these questions in a PM:

Why is having the bias diode and temp comp transistor coupled to the outputs bad and having them coupled to the drivers better?
Wouldn't replacing H760 with a transistor of higher hFE reduce the bias?

I think they are pertinent questions and I would like to take some time to explain the situation. For this, I considered a simplified analysis that you can consult in the jointed file.

For this analysis I made the assumption that the Vbe of transistors is not affected by the collector current. It is an acceptable assumption if we consider that the bias current is low. I made the assumption that the Vbe of transistors is only affected by their respective temperature, at a rate of -2mV/C.

For the first step (page 1), I computed the voltages and currents at room temperature for a simplified circuit of the 2270. Take note that this analysis is not valid for serial number below 3901 because there was major modifications in the output stage between SN 3900 and 3901.

Assuming 600mV for the Vbe of each transistor at room temperature, the bias voltage source (Vce of H760) must be adjusted to 1257mV in order to have 5mV between test points J654 and J756. This corresponds to 25mA in R788, that is 6mA from H758 and 19mA from H001.

What would happen to the bias current in H001 if only the output transistors heat up in temperature, and the bias voltage source is not thermally compensated. For this situation (Case 1, page 2) we still have a bias voltage source of 1257mV, and 600mV for the Vbe of H758 and H759. However, the Vbe of H001 and H002 is reduced to 560mV (for a temperature rise of 20 degrees C). The final result of this is a rising in the collector current of H001 from 19mA to 27mA.

Now, what would happen to the bias current in H001 if only the driver transistors heat up in temperature, and the bias voltage source is not thermally compensated. For this situation (Case 2, page 3) we still have a bias voltage source of 1257mV, and 600mV for the Vbe of H001 and H002. However, the Vbe of H758 and H759 is reduced to 560mV (for the same temperature rise of 20 degrees C). The final result of this is a rising in the collector current of H001 from 19mA to 219mA!!!

We can see that the bias current in the output transistors is much more affected by variations in temperature of the driver transistors than it is from variations in temperature of the power transistors. In order to maintain a constant bias current in H001 and H002, we must reduce the bias voltage source in the same proportion as the reduction of the Vbe of H758 and H759. An easy solution is to fix H761 to H758, and fix H760 to H759.

The stability of the bias current in H001 and H002 can be significantly improved, in regards to variations of temperature of H001 and H002, if R784 and R785 are reduced. If they are replaced by jumpers, the bias current in H001 and H002 would be unafected by variations in temperature of H001 and H002. However, this bias current would be very large if the forward voltage drop in H761 and the Vbe of H760 are larger than the Vbe of H758 and H759. This is the actual situation in Randy's amp and this is why I suggested to replace R784 and R785 by larger value. But if the thermally sensitive biasing components are fixed to the driver transistors, these resistors can probably be replace by jumpers. I think so because the driver transistors dissipate approximately 300mW and they are probably warmer than the output transistors.

 

[westend]

My respecto-meter just pegged!:thmbsp::thmbsp::thmbsp:

 

[dr*audio]

After my previous post I was asked these questions in a PM:

Why is having the bias diode and temp comp transistor coupled to the outputs bad and having them coupled to the drivers better?
Wouldn't replacing H760 with a transistor of higher hFE reduce the bias?

I think they are pertinent questions and I would like to take some time to explain the situation. For this, I considered a simplified analysis that you can consult in the jointed file.

For this analysis I made the assumption that the Vbe of transistors is not affected by the collector current. It is an acceptable assumption if we consider that the bias current is low. I made the assumption that the Vbe of transistors is only affected by their respective temperature, at a rate of -2mV/C.

For the first step (page 1), I computed the voltages and currents at room temperature for a simplified circuit of the 2270. Take note that this analysis is not valid for serial number below 3901 because there was major modifications in the output stage between SN 3900 and 3901.

Assuming 600mV for the Vbe of each transistor at room temperature, the bias voltage source (Vce of H760) must be adjusted to 1257mV in order to have 5mV between test points J654 and J756. This corresponds to 25mA in R788, that is 6mA from H758 and 19mA from H001.

What would happen to the bias current in H001 if only the output transistors heat up in temperature, and the bias voltage source is not thermally compensated. For this situation (Case 1, page 2) we still have a bias voltage source of 1257mV, and 600mV for the Vbe of H758 and H759. However, the Vbe of H001 and H002 is reduced to 560mV (for a temperature rise of 20 degrees C). The final result of this is a rising in the collector current of H001 from 19mA to 27mA.

Now, what would happen to the bias current in H001 if only the driver transistors heat up in temperature, and the bias voltage source is not thermally compensated. For this situation (Case 2, page 3) we still have a bias voltage source of 1257mV, and 600mV for the Vbe of H001 and H002. However, the Vbe of H758 and H759 is reduced to 560mV (for the same temperature rise of 20 degrees C). The final result of this is a rising in the collector current of H001 from 19mA to 219mA!!!

We can see that the bias current in the output transistors is much more affected by variations in temperature of the driver transistors than it is from variations in temperature of the power transistors. In order to maintain a constant bias current in H001 and H002, we must reduce the bias voltage source in the same proportion as the reduction of the Vbe of H758 and H759. An easy solution is to fix H761 to H758, and fix H760 to H759.

The stability of the bias current in H001 and H002 can be significantly improved, in regards to variations of temperature of H001 and H002, if R784 and R785 are reduced. If they are replaced by jumpers, the bias current in H001 and H002 would be unafected by variations in temperature of H001 and H002. However, this bias current would be very large if the forward voltage drop in H761 and the Vbe of H760 are larger than the Vbe of H758 and H759. This is the actual situation in Randy's amp and this is why I suggested to replace R784 and R785 by larger value. But if the thermally sensitive biasing components are fixed to the driver transistors, these resistors can probably be replace by jumpers. I think so because the driver transistors dissipate approximately 300mW and they are probably warmer than the output transistors.

This analysis makes a lot of sense. In practice, the only amp I know of where the bias transistor is mounted to the driver transistors, the Sansui AU-9900 and AU-11000; the output transistor current is very difficult to stabilize. I found the only way to set it properly is to set it and then monitor it with the cabinet in place, re-adjusting as necessary. I think this may be because the drivers and bias transistor are mounted on a good size heatsink, and may not get hot enough to affect the bias transistor.I always felt that if the bias transistor was mounted on the heatsink with the outputs, the idle current would be better controlled. I will have to try this to see if it helps.

 

[rBuckner]

Had a similar problem once with a 4300 receiver (similar output design) (about 25 years ago so YMMV) Couldn't get the bias down in one channel. Apparently Marantz designs the circuit with low tolerance to transistor variation. Turned out to be the VBE drop of the replacement drivers (not the outputs) was about 100 mv less than the original parts. I simply measured the b-e junctions of a couple of different subs, and used the ones that measured highest and everything worked fine. Hope this helps. Steve

Thanks! I did notice the Vbe drop on the new outputs was about 50mV different and wrote it down at home. Naturally I'm at work now so can't say whether it was more or less than the originals. I may just try to do the Vbe matching idea. Again, thanks for helping out.

 

[rBuckner]

Increased Vbe drop would (should) definitely help here, without other circuit modifications.

Shorting that driver bases together styill produced a voltage drop across an emmiter resistor(s) though. Something's weird about that.:scratch2:

Agreed on the weird part! As you will see on the posts after yours is some good info to chew on.

 

[rBuckner]

This analysis makes a lot of sense. In practice, the only amp I know of where the bias transistor is mounted to the driver transistors, the Sansui AU-9900 and AU-11000; the output transistor current is very difficult to stabilize. I found the only way to set it properly is to set it and then monitor it with the cabinet in place, re-adjusting as necessary. I think this may be because the drivers and bias transistor are mounted on a good size heatsink, and may not get hot enough to affect the bias transistor.I always felt that if the bias transistor was mounted on the heatsink with the outputs, the idle current would be better controlled. I will have to try this to see if it helps.

Yes it does. I already figured that it works for the bias transistor to be mounted to the heatsink as it has on all the units I've worked on it must be okay and I have some other issue. Post results if you try this out. I'll be updating this thread when I get back on this case.

 

[rBuckner]

After my previous post I was asked these questions in a PM:

Why is having the bias diode and temp comp transistor coupled to the outputs bad and having them coupled to the drivers better?
Wouldn't replacing H760 with a transistor of higher hFE reduce the bias?

I think they are pertinent questions and I would like to take some time to explain the situation. For this, I considered a simplified analysis that you can consult in the jointed file.

For this analysis I made the assumption that the Vbe of transistors is not affected by the collector current. It is an acceptable assumption if we consider that the bias current is low. I made the assumption that the Vbe of transistors is only affected by their respective temperature, at a rate of -2mV/C.

For the first step (page 1), I computed the voltages and currents at room temperature for a simplified circuit of the 2270. Take note that this analysis is not valid for serial number below 3901 because there was major modifications in the output stage between SN 3900 and 3901.

Assuming 600mV for the Vbe of each transistor at room temperature, the bias voltage source (Vce of H760) must be adjusted to 1257mV in order to have 5mV between test points J654 and J756. This corresponds to 25mA in R788, that is 6mA from H758 and 19mA from H001.

What would happen to the bias current in H001 if only the output transistors heat up in temperature, and the bias voltage source is not thermally compensated. For this situation (Case 1, page 2) we still have a bias voltage source of 1257mV, and 600mV for the Vbe of H758 and H759. However, the Vbe of H001 and H002 is reduced to 560mV (for a temperature rise of 20 degrees C). The final result of this is a rising in the collector current of H001 from 19mA to 27mA.

Now, what would happen to the bias current in H001 if only the driver transistors heat up in temperature, and the bias voltage source is not thermally compensated. For this situation (Case 2, page 3) we still have a bias voltage source of 1257mV, and 600mV for the Vbe of H001 and H002. However, the Vbe of H758 and H759 is reduced to 560mV (for the same temperature rise of 20 degrees C). The final result of this is a rising in the collector current of H001 from 19mA to 219mA!!!

We can see that the bias current in the output transistors is much more affected by variations in temperature of the driver transistors than it is from variations in temperature of the power transistors. In order to maintain a constant bias current in H001 and H002, we must reduce the bias voltage source in the same proportion as the reduction of the Vbe of H758 and H759. An easy solution is to fix H761 to H758, and fix H760 to H759.

The stability of the bias current in H001 and H002 can be significantly improved, in regards to variations of temperature of H001 and H002, if R784 and R785 are reduced. If they are replaced by jumpers, the bias current in H001 and H002 would be unafected by variations in temperature of H001 and H002. However, this bias current would be very large if the forward voltage drop in H761 and the Vbe of H760 are larger than the Vbe of H758 and H759. This is the actual situation in Randy's amp and this is why I suggested to replace R784 and R785 by larger value. But if the thermally sensitive biasing components are fixed to the driver transistors, these resistors can probably be replace by jumpers. I think so because the driver transistors dissipate approximately 300mW and they are probably warmer than the output transistors.

Thanks for going through the mental gymnastics! I have printed this out for reference when I dig back into this one. I'll be smarter when this is all figured out and fixed. :nerd:

 

[catrafter]

Thanks ecluser!
I continue to learn from your posts!
Tom

 

[ecluser]

In my last post I forgot to talk about the second question:

Wouldn't replacing H760 with a transistor of higher hFE reduce the bias?

The key factor is not the hFE of H760, but the Vbe. At any time, the bias voltage is the sum of the Vbe of H760 and the forward voltage drop (Vf) of H761. In order to reduce the bias voltage we must find components with a lower voltage drop.

For H760 you may try the 2SD669. This transitor has the lowest Vbe in a TO-126 body of any transistor that I know, typically less than 600mV at collector current as high as 25mA.

For H761 you may consider to use a diode connected (B and C connected together)transistor. The 2SD669 is again a good candidate.

But it is not easy to trim this circuit. Don't change both or your bias voltage will be too low...

The best option would be to modify the bias circuit in a true Vbe multiplier. This way you can adjust the bias voltage at any value you want. The modification consist of replacing H761 by a resistor of approximately the same value as the resistor connected between the base and emitter of H760. For ease of adjustment, you must add a resistor in series with the trim pot, approximately 5 to 10 times larger than the trim pot. For example you may replace H761 by a 8k2 resistor and add a 7k5 resistor in series with the actual 1k trim pot. For the initial trial you must preset the trim pot at his highest resistance.

 

[rBuckner]

In my last post I forgot to talk about the second question:

Wouldn't replacing H760 with a transistor of higher hFE reduce the bias?

The key factor is not the hFE of H760, but the Vbe. At any time, the bias voltage is the sum of the Vbe of H760 and the forward voltage drop (Vf) of H761. In order to reduce the bias voltage we must find components with a lower voltage drop.

For H760 you may try the 2SD669. This transitor has the lowest Vbe in a TO-126 body of any transistor that I know, typically less than 600mV at collector current as high as 25mA.

For H761 you may consider to use a diode connected (B and C connected together)transistor. The 2SD669 is again a good candidate.

But it is not easy to trim this circuit. Don't change both or your bias voltage will be too low...

The best option would be to modify the bias circuit in a true Vbe multiplier. This way you can adjust the bias voltage at any value you want. The modification consist of replacing H761 by a resistor of approximately the same value as the resistor connected between the base and emitter of H760. For ease of adjustment, you must add a resistor in series with the trim pot, approximately 5 to 10 times larger than the trim pot. For example you may replace H761 by a 8k2 resistor and add a 7k5 resistor in series with the actual 1k trim pot. For the initial trial you must preset the trim pot at his highest resistance.

I will be ordering stuff by Monday for sure anyway, I'll see if the D669 is available along with the other stuff. Yes, one thing at a time will be tried. I really want to know EXACTLY what's going on. I may just be able to sneak some time in on it this weekend... got my fingers crossed! Thanks so much ecluser, you should write a book on fixing amp circuits and sell it on AK. Really.