Good for all, I am a follower of this site, but is the first time I need your help.
I have an integrated amplifier Rotel RA-970BX, in immaculate condition and in apparent good performance. But out of curiosity I measured with a multimeter at the terminals of speakers and I find that has left channel DC offset at least 45mv and more 68mv right channel.
Measurements were performed with the amplifier with no load, with zero volume and after 10 minutes of ignition.
I have understood that an amplifier should not exceed the 15mV offset. I'd like to target the solution. I am told that the transistors 2sa1016 and 2sc2362 have been old and must be replaced. I should also replace the electrolytic capacitors in the power stage. thank to help me. Greetings Jose Alpuy (From Uruguay. Here there is no technical service Rotel).
I have an integrated amplifier Rotel RA-970BX, in immaculate condition and in apparent good performance. But out of curiosity I measured with a multimeter at the terminals of speakers and I find that has left channel DC offset at least 45mv and more 68mv right channel.
Measurements were performed with the amplifier with no load, with zero volume and after 10 minutes of ignition.
I have understood that an amplifier should not exceed the 15mV offset. I'd like to target the solution. I am told that the transistors 2sa1016 and 2sc2362 have been old and must be replaced. I should also replace the electrolytic capacitors in the power stage. thank to help me. Greetings Jose Alpuy (From Uruguay. Here there is no technical service Rotel).
Somebody else on here recently had the same problem with the same amp I think. The conclusion was that it is inherent to the design and not a problem.
The thread starter had been alarmed by a website saying that the offset was a problem and caused by drifting transistors, which should be changed. Both of these statements are completely untrue.
The thread starter had been alarmed by a website saying that the offset was a problem and caused by drifting transistors, which should be changed. Both of these statements are completely untrue.
I appreciate your opinions. I've bought on eBay matched transistors and capacitors Nichicon audio type (I'm still waiting for them). My intention is to change, but would not if any of you tell me that would be detrimental.
I am very pleased with the quick response I've had in this forum.
I would like to display photos and comments from a project I've done, taking the cabinet of a Technics SA-5551 where I placed LM3886 amps, preamps and toroidal transformer and phono line of Rod Elliott circuits using OPA2134, but still not as post . Regards, Jose Alpuy.
I am very pleased with the quick response I've had in this forum.
I would like to display photos and comments from a project I've done, taking the cabinet of a Technics SA-5551 where I placed LM3886 amps, preamps and toroidal transformer and phono line of Rod Elliott circuits using OPA2134, but still not as post . Regards, Jose Alpuy.
I would not expect changing of capacitors to change the output offset.
Have you shorted the input and left the output with out a load to measure the output offset? If not then do it again, properly.
If the NFB capacitor is leaking badly, this will change the output offset. I think the only way to check this is to remove the NFB cap and reform it and find out what leakage it is passing. If it fails to meet specification then replace it with a freshly reformed capacitor that has been fully discharged.
Have you shorted the input and left the output with out a load to measure the output offset? If not then do it again, properly.
If the NFB capacitor is leaking badly, this will change the output offset. I think the only way to check this is to remove the NFB cap and reform it and find out what leakage it is passing. If it fails to meet specification then replace it with a freshly reformed capacitor that has been fully discharged.
The only effective offset control is a DC servo. Any passive adjustment is just as richie said.
OK, your tweeter has a series cap. No problem. If you have a 3 way, your mid has a blocking cap. So it is only your woofer you are worried about. They are very tough. Assume a 7 Ohm Re. .065V is only .6mW
Go have a pint and find something else to worry about.
OK, your tweeter has a series cap. No problem. If you have a 3 way, your mid has a blocking cap. So it is only your woofer you are worried about. They are very tough. Assume a 7 Ohm Re. .065V is only .6mW
Go have a pint and find something else to worry about.
Thank you very much. Now I'm calmer, but I worried about damage to my B & W dm602. If there is damage, here in Uruguay impossible to replace drivers.
Because I have asked eBay to recap and matched transistors of the differential input, try to replace them when they arrive, and if no change, to drink beer and enjoy the good music! even with "DC offset". I have been very pleased with all your contributions.
I'll find a question that formularles, because I am a SOLIDPHONO constuyendo TNT (which I think is a clone of Lehman Black cube). Greetings and thank you very much. José.
Because I have asked eBay to recap and matched transistors of the differential input, try to replace them when they arrive, and if no change, to drink beer and enjoy the good music! even with "DC offset". I have been very pleased with all your contributions.
I'll find a question that formularles, because I am a SOLIDPHONO constuyendo TNT (which I think is a clone of Lehman Black cube). Greetings and thank you very much. José.
Davidsrsb Dear, I have some experience in assembling audio circuits. I have identified in the service manual of RA_970BX Rotel components have suggested I renew. What I understood is that if you tell me how to do it, do it, and if not to stop as it is?. sometimes happens that the English give me some confusion. So I are saying all 68mv offset is not as dangerous. I just want this amp best yield can be achieved. Regards, Jose.
It looks like the folks at Rotel did not bother to match DC resistance, resulting in this higher-than-usual output offset (which, as mentioned, still is by no means dangerous). If you do want to reduce it further, swap R683 and R684 (47k) for, say, 22k. That should bring it down to about 1/3 of what it is now.
There is one area where you could improve performance quite a bit if you feel comfortable swapping out an opamp: Background noise. The Rotel folks had a penchant for fast but notoriously noisy opamps after the volume control - an AD827JN in this case (en = 15 nV/sqrt(Hz)), IC501. You could install an LM4562 (or one of its LME49xxx relatives) in its place; if you cannot find one, the venerable NE5532 should also do.
There is one area where you could improve performance quite a bit if you feel comfortable swapping out an opamp: Background noise. The Rotel folks had a penchant for fast but notoriously noisy opamps after the volume control - an AD827JN in this case (en = 15 nV/sqrt(Hz)), IC501. You could install an LM4562 (or one of its LME49xxx relatives) in its place; if you cannot find one, the venerable NE5532 should also do.
I have 3 of these amps, one modified but the other 2 waiting to be modified, the worst one has 24 mv offset on one channel, these amps are nearly 20 years old. They use blackgates which I doudbt have problems as yet. That said the amp with the large offset has had one output transistor replaced hence the larger offset.Might be the case here too, original rotels have matched output pairs from factory.
sgrossklass, I totally disagree with you, the ad827 is the best sounding opamp for this amp. The noise is so low you cant hear anything, so why mess with it, in my view its a downgrade and many designers will tell you that, I wont even mention how much of a downgrade in case of the 5532.
sgrossklass, I totally disagree with you, the ad827 is the best sounding opamp for this amp. The noise is so low you cant hear anything, so why mess with it, in my view its a downgrade and many designers will tell you that, I wont even mention how much of a downgrade in case of the 5532.
Output transistors have nothing to do with it, it's the inputs. In theory, the npn and pnp inputs (in this dual differential input design) would be exactly complementary and running at exactly the same current - in practice, they're not, and those transistors also exhibit aging to some degree, causing leakage to creep up over time. Even if the transistors in each differential pair were aging equally (there's a significant statistical component involved), npns and pnps would still differ. I'd guess the pnps are more critical.
I am not sure what kind of architecture the AD827/847 uses - I'm guessing an AB push-pull type à la OP467? Or folded cascode like AD826 and AD829? Either way, input impedance nonlinearity should be reasonably low, and transfer linearity more than adequate for the purpose at least. It shouldn't be bothered by high frequencies all that much, nor will capacitive loading upset it according to the datasheet. So it looks like a good match for the circuit here, which upon closer inspection is not too generous with high-frequency input filtering nor does it provide series resistors for the opamp outputs (cheapskates...). Except, well, it's noisy, which is not a good match for a circuit with all its gain after the volume pot.
Calculated output noise is about 330 µVrms @ 20 kHz b/w, which is decidedly average. This kind of noise level produces a clearly audible - if not yet objectionable - noise floor at 1 m from speakers rated at 88 dB / 2.83 V, which is hardly Klipschorn-level sensitivity. (I'm in 4-ohm land, which on average is 3 dB more sensitive.) It is not that uncommon for mass-market integrated amps to exhibit a residual noise floor of 100..150 µV (most any modern-day Denon, Marantz, NAD and the little Pioneers). Even PGA-equipped gear seems to have gotten fairly good if you look at the specs of the latest stereo receivers from Yamaha or Onkyo.
Considering the potential issue of capacitive loading, a TI NE5532 may not be all that happy indeed (a case for a scope). The LM4562 is supposed to drive up to 100 pF at least. Adding 100 ohms in series with coupling caps C505 and C506 (10µ/50 - yeah, it may not look pretty) should get this sorted in any case. At least supply bypassing seems to be good.
I am not sure what kind of architecture the AD827/847 uses - I'm guessing an AB push-pull type à la OP467? Or folded cascode like AD826 and AD829? Either way, input impedance nonlinearity should be reasonably low, and transfer linearity more than adequate for the purpose at least. It shouldn't be bothered by high frequencies all that much, nor will capacitive loading upset it according to the datasheet. So it looks like a good match for the circuit here, which upon closer inspection is not too generous with high-frequency input filtering nor does it provide series resistors for the opamp outputs (cheapskates...). Except, well, it's noisy, which is not a good match for a circuit with all its gain after the volume pot.
Calculated output noise is about 330 µVrms @ 20 kHz b/w, which is decidedly average. This kind of noise level produces a clearly audible - if not yet objectionable - noise floor at 1 m from speakers rated at 88 dB / 2.83 V, which is hardly Klipschorn-level sensitivity. (I'm in 4-ohm land, which on average is 3 dB more sensitive.) It is not that uncommon for mass-market integrated amps to exhibit a residual noise floor of 100..150 µV (most any modern-day Denon, Marantz, NAD and the little Pioneers). Even PGA-equipped gear seems to have gotten fairly good if you look at the specs of the latest stereo receivers from Yamaha or Onkyo.
Considering the potential issue of capacitive loading, a TI NE5532 may not be all that happy indeed (a case for a scope). The LM4562 is supposed to drive up to 100 pF at least. Adding 100 ohms in series with coupling caps C505 and C506 (10µ/50 - yeah, it may not look pretty) should get this sorted in any case. At least supply bypassing seems to be good.
Last edited:
Looking at the schematic there s mistakes that if implemented
could only be dreadfull...
The pre drivers have their bases connected to their collectors.
The PNP pre driver and driver have no emitter resistors
and the common point with the emittor resistors of the
complementary pre driver and drivers is connected not to
the output but directly to the inverting input....
The feedback signal is sampled after the output fuse but
this was surely an attempt to reduce the fuse induced distorsion
by inserting it in the FB loop.
As for the DC output offset , as pointed by Sgrossklass , it is highly
dependant of the input differentials matching , the most important being
that the NPN pair must be matched in matter of Vbe , and the same
for the PNP pair.
Matching of the NPN pair in respect of the PNP pair is less
important , what matters is that each pair is matched.
Last but not least , the input biaising resistor (R603/604) has an
excess value of 35.5K , that is, a bias current in the range of 2uA induce
as much as 71mV output DC shift compared to the value that it would
have for a theorical optimal resistor value, i.e , 12K.
Better is to reduce it to 22K first , make a measurement
and if not enough reducing it further.
could only be dreadfull...
The pre drivers have their bases connected to their collectors.
The PNP pre driver and driver have no emitter resistors
and the common point with the emittor resistors of the
complementary pre driver and drivers is connected not to
the output but directly to the inverting input....
The feedback signal is sampled after the output fuse but
this was surely an attempt to reduce the fuse induced distorsion
by inserting it in the FB loop.
As for the DC output offset , as pointed by Sgrossklass , it is highly
dependant of the input differentials matching , the most important being
that the NPN pair must be matched in matter of Vbe , and the same
for the PNP pair.
Matching of the NPN pair in respect of the PNP pair is less
important , what matters is that each pair is matched.
Last but not least , the input biaising resistor (R603/604) has an
excess value of 35.5K , that is, a bias current in the range of 2uA induce
as much as 71mV output DC shift compared to the value that it would
have for a theorical optimal resistor value, i.e , 12K.
Better is to reduce it to 22K first , make a measurement
and if not enough reducing it further.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.