The advantages of stage 5 are probably twofold:
1. The obvious effect of the current mirror, though there would still be a small error from follower input current.
2. While fiddling with the current mirror, you "accidentally" made the VAS considerably more robust against nonlinear loading (as provided by the output stage under load) by inserting the follower. Samuel Groner discussed this in his notes of Douglas Self's power amplifier book (along with other interesting topics, well worth reading).
You might be able to get the 2nd harmonic down a bit further by replacing the follower emitter resistor by a current sink, at least I found in simulation that this would eliminate the propensity for common-mode distortion of this VAS topology. It's more of an issue for a general purpose opamp than for a speaker power amp with its typically high gain, but anyway. Other than that it looks like you're already hitting the limits of parts tolerances on the even and output stage distortion / loop gain on the odd harmonics. Not like there'd be much of anything left to complain about at this level, Douglas Self doesn't get more than about 6 dB better than that either if memory serves.
So where are those still lurking? Hmm. Have you tried dropping R635? Generally speaking, added series R inside the feedback loop usually makes performance and stability worse rather than better. In this case it might help with the nonlinearity of the fuse though (the presence of which arguably is more useful than distortion below -100 dB).
It would be interesting to know what THD(f) looks like. The U7 can record at up to 192 kHz so I'd expect decent results to about 10 kHz at least. Thermal issues (e.g. from the fuse) should show up in the bass.
As far as the TL011C goes, I know it sounds stupid but are you absolutely, 100% certain that you had the input and output the right way around? Other than that one would have to test whether the parts are working at all.
1. The obvious effect of the current mirror, though there would still be a small error from follower input current.
2. While fiddling with the current mirror, you "accidentally" made the VAS considerably more robust against nonlinear loading (as provided by the output stage under load) by inserting the follower. Samuel Groner discussed this in his notes of Douglas Self's power amplifier book (along with other interesting topics, well worth reading).
You might be able to get the 2nd harmonic down a bit further by replacing the follower emitter resistor by a current sink, at least I found in simulation that this would eliminate the propensity for common-mode distortion of this VAS topology. It's more of an issue for a general purpose opamp than for a speaker power amp with its typically high gain, but anyway. Other than that it looks like you're already hitting the limits of parts tolerances on the even and output stage distortion / loop gain on the odd harmonics. Not like there'd be much of anything left to complain about at this level, Douglas Self doesn't get more than about 6 dB better than that either if memory serves.
So where are those still lurking? Hmm. Have you tried dropping R635? Generally speaking, added series R inside the feedback loop usually makes performance and stability worse rather than better. In this case it might help with the nonlinearity of the fuse though (the presence of which arguably is more useful than distortion below -100 dB).
It would be interesting to know what THD(f) looks like. The U7 can record at up to 192 kHz so I'd expect decent results to about 10 kHz at least. Thermal issues (e.g. from the fuse) should show up in the bass.
As far as the TL011C goes, I know it sounds stupid but are you absolutely, 100% certain that you had the input and output the right way around? Other than that one would have to test whether the parts are working at all.
As far as the TL011C goes, I know it sounds stupid but are you absolutely, 100% certain that you had the input and output the right way around? Other than that one would have to test whether the parts are working at all.
p { margin-bottom: 0.25cm; line-height: 120%; }a:link { } He, he sgrossklass. You know, I did consider that possibility as I have on more than one occasion found myself to be a absolute plonker in that respect. But unless the datasheet pinout is wrong, it was connected correctly.
So, I set up a quick test system with a bench power supply (1mA current limited) sourcing two hi-bright LED's into the two branches of the TL011C current mirror. Slowly increasing the voltage, the two LED's lit up at the same rate and intensity, so the current mirror seems to work - with about 1.4V across both input and output branch. That should be ok in the amp circuit, hmmm.... maybe I just gave up too soon?
Anyway, disaster struk a day later. I was measuring the THD of the two-stage Miller network “Mod 6” when I noticed that the spectrum trace occasionally started to re-set its averaging at strange intervals. First I thought that it could be some RTA software interrupt thing, but then I heard a very faint click from the speaker protection relays when the trace re-set. Hello, an intermittently faulty component somewhere?
So I got out my trusted IR temp measuring device and set the amp to into stress test mode; pink noise input / full whack power output into the dummy resistor load - while I constantly checked the temperature rise of all transistors. Yes, the output power devices got warm (as did the load), then very warm (65 oC) but nothing seemed blatantly untoward, until – a final click and all lights out.
The transformer's 230V primary winding had gone to the eternal Milli-Henry hunting grounds.
Blast!
Ouch. The poor little transformer obviously wasn't built with hardcore load testing in mind, so the thermal fuse ultimately wasn't amused.
The difficulty of thermal fuse replacement seems to vary significantly, no idea how it is with these Rotels. One may have to borrow one from another unit. Parts quality in this model seems to have been quite decent so I'd expect the same from the xmfr. Of course if you happen to have a super high efficiency part of similar voltages and power ratings floating around, that may be worth a shot. I wouldn't want to use a super oversized one just in case voltage drop was used to keep the output transistors in SOA.
The difficulty of thermal fuse replacement seems to vary significantly, no idea how it is with these Rotels. One may have to borrow one from another unit. Parts quality in this model seems to have been quite decent so I'd expect the same from the xmfr. Of course if you happen to have a super high efficiency part of similar voltages and power ratings floating around, that may be worth a shot. I wouldn't want to use a super oversized one just in case voltage drop was used to keep the output transistors in SOA.
Clinical update:
A successful emergency Transformerectomy was performed on a well-known ailing RA-820AX this morning. The patient is still a bit unresponsive, but is otherwise doing as well as could be expected under the current circumstances.
Further, a suitable replacement part was received by courier later in the day and is now being prepared for a transplantation scheduled as soon as the operating table is cleared from other minor procedures.
AngelP
Consultant Audio Surgeon, MD, THR
(Mega Dork, Tin Hat Required)
A successful emergency Transformerectomy was performed on a well-known ailing RA-820AX this morning. The patient is still a bit unresponsive, but is otherwise doing as well as could be expected under the current circumstances.
Further, a suitable replacement part was received by courier later in the day and is now being prepared for a transplantation scheduled as soon as the operating table is cleared from other minor procedures.
AngelP
Consultant Audio Surgeon, MD, THR
(Mega Dork, Tin Hat Required)
Hi AngelP,
You did this the right way. Thank you so much for this example for others! How much did the new transformer cost (if you don't mind me asking)?
Some times you can gain access to the thermal fuse and replace it. I'm always afraid someone will just short it as I have had equipment come in for service with a shorted thermal fuse and a very burnt transformer. That fuse is necessary. The laws in your area may or may not address that issue, but if you can get the exact same fuse and replace the burnt one, you should be okay. Without this fuse, a fault in the unit can cause a fire - a really messy, stinking fire that will write-off the equipment.
I only mention this because someone will usually suggest to short the fuse. Please, never even consider doing that. I was able to get replacement fuses for some Nakamichi products a long time ago. Today might be a different situation completely. I do know that if you mess with the primary circuit in equipment, your insurance will probably not cover fire damage at all. Is the risk worth this (no!, in case someone has to think about it)?
-Chris
You did this the right way. Thank you so much for this example for others! How much did the new transformer cost (if you don't mind me asking)?
Some times you can gain access to the thermal fuse and replace it. I'm always afraid someone will just short it as I have had equipment come in for service with a shorted thermal fuse and a very burnt transformer. That fuse is necessary. The laws in your area may or may not address that issue, but if you can get the exact same fuse and replace the burnt one, you should be okay. Without this fuse, a fault in the unit can cause a fire - a really messy, stinking fire that will write-off the equipment.
I only mention this because someone will usually suggest to short the fuse. Please, never even consider doing that. I was able to get replacement fuses for some Nakamichi products a long time ago. Today might be a different situation completely. I do know that if you mess with the primary circuit in equipment, your insurance will probably not cover fire damage at all. Is the risk worth this (no!, in case someone has to think about it)?
-Chris
Chris,
I totally agree and would never short a fuse. Finding a proper Rotel thermal fuse replacement would probably be a nightmare. In any case, the old Rotel transformer is built like a tank with welded steel cover and soldered copper belly band. It would simply not be worth the time and effort to go digging after the thermal fuse and the end result could well look like a dog's breakfast. So it is off to recycling.
But I now got the opportunity to see if a toroidal would reduce the stubborn 50Hz mains noise, so I found a well built 2 x 25Vac 160VA from Maplin for only GBP 25 - with free delivery! Funny that, I seem to remember Maplin as being one of the more pricey retailers - but I guess that things change.
For my own peace of mind I was also considering to put in a primary fuse. Ok, the UK mains plugs are fused, but some with 5A, other amp plugs come in with 13A fuses put in. Not much hot transformer protection in that.
sgrossklass,
Thank you for the link to the paper by Samuel Groner, I found that very interesting indeed.
For your question regarding THD(F), would you really think that there would be any quantifiable difference to THD(R) at these very low distortion levels?
And yes, I did have a second look at the unusual R365 (the 0.22ohm resistor in series with the output and fuse, but within the feedback loop). I can't see why its presence (or not) should cause any stability issues, but if oscillation should ever show its ugly head
, this design would allow for a small pF cap to be put in between the amp output and the non-inverting input to add another pole and hopefully take care of that?
Per
I totally agree and would never short a fuse. Finding a proper Rotel thermal fuse replacement would probably be a nightmare. In any case, the old Rotel transformer is built like a tank with welded steel cover and soldered copper belly band. It would simply not be worth the time and effort to go digging after the thermal fuse and the end result could well look like a dog's breakfast. So it is off to recycling.
But I now got the opportunity to see if a toroidal would reduce the stubborn 50Hz mains noise, so I found a well built 2 x 25Vac 160VA from Maplin for only GBP 25 - with free delivery! Funny that, I seem to remember Maplin as being one of the more pricey retailers - but I guess that things change.
For my own peace of mind I was also considering to put in a primary fuse. Ok, the UK mains plugs are fused, but some with 5A, other amp plugs come in with 13A fuses put in. Not much hot transformer protection in that.
sgrossklass,
Thank you for the link to the paper by Samuel Groner, I found that very interesting indeed.
For your question regarding THD(F), would you really think that there would be any quantifiable difference to THD(R) at these very low distortion levels?
And yes, I did have a second look at the unusual R365 (the 0.22ohm resistor in series with the output and fuse, but within the feedback loop). I can't see why its presence (or not) should cause any stability issues, but if oscillation should ever show its ugly head
, this design would allow for a small pF cap to be put in between the amp output and the non-inverting input to add another pole and hopefully take care of that?Per
Hi Per,
With the new transformer being that inexpensive, it is the only rational way to go. I think you have great luck!
Most manufacturers do not make the thermal fuses available as spares. That is a good thing in many ways. Nakamichi only did this during a time when there was an issue with Canadian core temperature cut-out being 10 °C lower than the UL fuse. At that time we began losing transformers at an amazing rate. 10 °C is a lot! It made the difference between open transformers and running ones.
-Chris
With the new transformer being that inexpensive, it is the only rational way to go. I think you have great luck!
Most manufacturers do not make the thermal fuses available as spares. That is a good thing in many ways. Nakamichi only did this during a time when there was an issue with Canadian core temperature cut-out being 10 °C lower than the UL fuse. At that time we began losing transformers at an amazing rate. 10 °C is a lot! It made the difference between open transformers and running ones.
-Chris
Blimey, Chris
I would have thought that you Canadians of all people needed all the heating you could get in your homes. Does that mean that class-A amps are banned in Canada?
Whenever it gets a bit nippy in my audio den, first I of course call to warn my energy supplier - then switch on the Hiraga le Monstre - and drift away to sweet Carribean reggae music and the blissful warmth from its red hot heatsinks. Yah, Man!
I would have thought that you Canadians of all people needed all the heating you could get in your homes. Does that mean that class-A amps are banned in Canada?
Whenever it gets a bit nippy in my audio den, first I of course call to warn my energy supplier - then switch on the Hiraga le Monstre - and drift away to sweet Carribean reggae music and the blissful warmth from its red hot heatsinks. Yah, Man!
Hi Per,
-Chris
Yes, we really do!
No, consider this. Gas heat is much less expensive than electric heat. People who use larger class A amplifiers have a self limiting feature built in. Their electricity bill! We call electricity, "Hydro" in this area. Water falls and dams are where most electrical power is created. That is excluding our very well behaved nuke facilities.
Canadians like Nelson Pass amplifiers. Distributed heat as we like whole house audio.
-Chris
Thanks AngelP!
I was only planning on a re-cap of my Rotel RA-820BX2, wich of course was humming very audibly, but having read this thread I couldn't help myself. I had do mod it.
I didn't want to omit the DC blocking input-capacitors C601/602, so I put 5.6µF MKPs in there (had them flying around) instead of the 1µF nude, shiny one of unknown construction. I also lowered the series resistors R601/602 from 1.8k to 1k and the parallel input one from 47k to 39k. C603/604 were 470pF!! (schematic calls for a still hefty 390pf) and are now 150pF.
Your "Stage 1" mods are now in there, along with a complete e-lythic recap of the amplifier, a new 35A rectifier, new (multi-turn) bias trimmers and a couple of .1% 25ppm metal film resistors instead of what I supposed were carbon film types (but actually are metal film types).
Strangely, the power capacitors were already 10,000µF ones (schematic says it should be 8,200s). But they looked stock, so out they went. Newer 10,000µF/50V/105°C types went in, instead.
I did not match the input pairs of A1016s or any capacitors between channels, but I did thicken some traces of the power supply. I would've loved to re-arrange that catastrophe of a grounding scheme so that the ripple of the large power supply caps doesn't interfere with the rest of the amp, but I ran out of f****s to give and mounted everything in as it was designed
The two secondaries got their .1µF MKP X2 filters. After Adjusting the bias to a still low and only luke-warm 6.5mV/30mA instead of the nearly pathetic 4.4mV/20mA the manual calls for, I called it a day.
Results:
Pros:
Since I'm not bound by any modification rules but am very much concerned about the stability of the amp - are a .1µF/10ohm zobel and a more or less random inductor that "looks the right size" a good idea, here? If so, what wattage resistor be enough in the zobel and inside the inductor?
Furthermore, I'd like to lower the gain of the amp. Sources nowdays are "a bit" more powerful than the 200mV standard 30 years ago. Would reducing the (in case of my RA-820BX2) 33K R637/638 feedback resistors to, say, 22K, do this trick or would I have to get an oscilloscope to check for stability?
The 100µF feedback-grounding caps are now 63V low-esr Panasonic FRs. Their specs say < 0.01 CV (μA) for DC-leakage and a tan alpha of 0.09; 5 times more leakage than a low leakage type - but still better than the 30 year old part it replaces. Is there even such a thing as a bipolar, low-leakage 100µF cap with radial leads? The axial ones used in speaker crossovers are "a bit" too big to fit into any amp without looking "too ghetto" for my tastes.
Curious detail: The phono section is made up of 4 virgin-made Signetics NE5534s composed of unobtanium, but their outputs are decoupled with tantalum caps
Thanks everyone for putting up with this unscientific wall of text...So many questions and so little knowledge don't go together silently
Any feedback would be greatly appreciated
I was only planning on a re-cap of my Rotel RA-820BX2, wich of course was humming very audibly, but having read this thread I couldn't help myself. I had do mod it.
I didn't want to omit the DC blocking input-capacitors C601/602, so I put 5.6µF MKPs in there (had them flying around) instead of the 1µF nude, shiny one of unknown construction. I also lowered the series resistors R601/602 from 1.8k to 1k and the parallel input one from 47k to 39k. C603/604 were 470pF!! (schematic calls for a still hefty 390pf) and are now 150pF.
Your "Stage 1" mods are now in there, along with a complete e-lythic recap of the amplifier, a new 35A rectifier, new (multi-turn) bias trimmers and a couple of .1% 25ppm metal film resistors instead of what I supposed were carbon film types (but actually are metal film types
).Strangely, the power capacitors were already 10,000µF ones (schematic says it should be 8,200s). But they looked stock, so out they went. Newer 10,000µF/50V/105°C types went in, instead.
I did not match the input pairs of A1016s or any capacitors between channels, but I did thicken some traces of the power supply. I would've loved to re-arrange that catastrophe of a grounding scheme so that the ripple of the large power supply caps doesn't interfere with the rest of the amp, but I ran out of f****s to give and mounted everything in as it was designed
The two secondaries got their .1µF MKP X2 filters. After Adjusting the bias to a still low and only luke-warm 6.5mV/30mA instead of the nearly pathetic 4.4mV/20mA the manual calls for, I called it a day.
Results:
Pros:
- Output DC down from 25/15 mV to 10/3 mV.
- Hum down from very much audible to only audible when amp is at full whack (something that rarely happens with the high-efficiency speakers it drives).
- sluggish, warm (and veiled) character of the Rotel is now much more analytical/neutral with more attack and control.
- The "thump" when switching the amp on is now intolerable
. I fear for my speakers....a 6.5€ speaker protection board is now on it's way to the rescue...from china.
Since I'm not bound by any modification rules but am very much concerned about the stability of the amp - are a .1µF/10ohm zobel and a more or less random inductor that "looks the right size" a good idea, here? If so, what wattage resistor be enough in the zobel and inside the inductor?
Furthermore, I'd like to lower the gain of the amp. Sources nowdays are "a bit" more powerful than the 200mV standard 30 years ago. Would reducing the (in case of my RA-820BX2) 33K R637/638 feedback resistors to, say, 22K, do this trick or would I have to get an oscilloscope to check for stability?
The 100µF feedback-grounding caps are now 63V low-esr Panasonic FRs. Their specs say < 0.01 CV (μA) for DC-leakage and a tan alpha of 0.09; 5 times more leakage than a low leakage type - but still better than the 30 year old part it replaces. Is there even such a thing as a bipolar, low-leakage 100µF cap with radial leads? The axial ones used in speaker crossovers are "a bit" too big to fit into any amp without looking "too ghetto" for my tastes.
Curious detail: The phono section is made up of 4 virgin-made Signetics NE5534s composed of unobtanium, but their outputs are decoupled with tantalum caps
Thanks everyone for putting up with this unscientific wall of text...So many questions and so little knowledge don't go together silently
Any feedback would be greatly appreciated
Hi 1210,
Multi-turn trimmers are potentiometers. The wiper is not designed to carry any current. Normal bias controls are rheostats and do pass current through the wiper. You can use new single turn controls, or the originals.
Before changing part types, learn about their characteristics first. Change-'em back!
* match the input diff pair and install the matched pairs so that they are in contact (each pair, not both pairs together).
* match the output transistors and drivers (can decrease THD by 20 dB before feedback).
-Chris
Okay, just do not play with it!
By doing this ....
Since my career has more than spanned that length of time, I can tell you with certainty that the consumer standard is and has always been approx. 316 mV (-10 dBu). What has changed is the peak to average levels in recordings from CDs. A CD has a much higher dynamic range than radio or cassette normally has. The old tape standard load of 10 K has risen to 47 K ( I had to include that optional load, 10 K, in my test jigs).
Excellent!
Please don't mess with compensation caps. You can cause the amplifier to become unstable. Replace those capacitors with the ones that were there, or something you feel might be better but the same values.
Same advice as above.
Multi-turn trimmers are potentiometers. The wiper is not designed to carry any current. Normal bias controls are rheostats and do pass current through the wiper. You can use new single turn controls, or the originals.
Before changing part types, learn about their characteristics first. Change-'em back!
This is the area where you will get the best bang for the buck. You either need a balancing jig, or you must ensure the transistors are measured at exactly the same temperature (very, very difficult to do). Now you know why I designed a jig for this.
I have to ask why you did this. The distortion climbs with added bias current past the sweet spot for one. A manufacturer will never select a bias current that is too low for the amplifier. Not ever, unless the heat sinks are too small and its getting too warm. I'm going to ask you very nicely to reduce the bias current back to the recommended levels. No. You did not hear any increase in quality as a result of increasing the bias current. The one thing that would really improve the distortion figure, you didn't do. If you want to actually improve the sound quality, there are two things you can do.
* match the input diff pair and install the matched pairs so that they are in contact (each pair, not both pairs together).
* match the output transistors and drivers (can decrease THD by 20 dB before feedback).
-Chris
Well...too late for that, now.
Unless I'm very much mistaken, the original 470pf cap gives the amp a -3dB point right around 21kHz at the input...at best. 150pF raises this to 66kHz. I might put a 300pF in there to lower this to 33kHz, but an input filter with a -3dB point inside the redbook 22kHz sounds like a lazy idea, that's why I changed it. Putting a 50k pot/stepped attenuator in there, the 150pF should give the desired >30kHz pole frequency at the input. I have no clue as to how this amp could call for a 40kHz -3dB point in its specs with this configuration. Magic, I guess.
How much current are we talking about, here? The standard single-turn pots, as noted pretty much everywhere I looked, are impossible to adjust decently. I will not put the originals back. That is, unless someone provides a mod so that the adjustment range is reasonable.
I did look into transistor matching and decided to either get matched pairs or just leave it be. I'm not going to make a test-jig to test tons of transistors (I also don't have...).
Bingo. The Rotels Heatsinks are 2 x 150x50x40mm small. It has tiny feet and half of one heatsink doesn't have any holes above it. Rotel correctly assumed that the average consumer will be an idiot that puts something on the amp. So, in order for it not to go up in smoke, it has a low bias. I will adress some of these issues and never, ever put something on top. Hence more bias. At the very least, I have very noisy central heating wich I hate to turn on - the amp heats my room more silently
I also happens to lower it's bias when temperature rises...
I didn't say I did. I just noted 20mA to be "on the low side of things" and raised it to 30mA. Just to be sure. The heatsink can take it - at least for the winter. But, thanks to you, it turned it back to 22mA. 10% over specs. Should be more than safe, given the non-functioning fine adjustability of the original pots. The transformers get warmer than the heatsink, now
Matched input pairs are still on the table, as is their propper mounting. Not shure about full-on Stage 2 or even Stage 5 (not knowing how to solder SMD). I've just begun refreshing the amp.
Matched outputs may be in the cards. It all depends on costs, me beeing a broke student. I don't want to sink 50€ into this amp just in matched transistors...only to discover that the volume knob is still not working...Regarding the volume pot and it's attenuation - any balance adjustment means there's always sound coming from some speaker, hence my "not enough attenuation"-comment.
So far, my to-do list:
- get matched input pairs, glue them together
- inductors + zobels on output (inside feedback loop or at speaker outs?)
- stepped 50k attenuator instead of current 20k volume pot
- mount a speaker protection board
- find the courage to cut traces to put an RC-filter into the supply of the input stage (as in Mod 2+)
I'd still be glad to get answers on the questions I have.
Last edited:
Hi 1210,
Congrats with the decision to venture into Mod territory! Enjoy the fun and the listening improvements!
Firstly, for all that follow this thread - please note that 1210 is modding a Rotel RA-820BX2 which is a quite different design to my Mod victim (RA-820AX).
Main differences are 1) No Line-in preamp in the BX2 (which is why the power amp section amplification is a whopping 37dB or 71 times) 2) Doubled output transistors and yet 3) only +/-31V power supplies. Also, there are some differences in component numbering, so 1210 - could you please attach a schematic for the BX2 so we know what we refer to? My copy is sort of smudgy-ish.
I will get back to you with my answers your questions, but the first thing that struck me from your description was: Has someone else been fiddling with the amp circuitry before you? If the answer is "yes" or "probably" then you will do well to first have a long and thorough check of all components and solder joints.
Cheers,
Per
Congrats with the decision to venture into Mod territory! Enjoy the fun and the listening improvements!
Firstly, for all that follow this thread - please note that 1210 is modding a Rotel RA-820BX2 which is a quite different design to my Mod victim (RA-820AX).
Main differences are 1) No Line-in preamp in the BX2 (which is why the power amp section amplification is a whopping 37dB or 71 times) 2) Doubled output transistors and yet 3) only +/-31V power supplies. Also, there are some differences in component numbering, so 1210 - could you please attach a schematic for the BX2 so we know what we refer to? My copy is sort of smudgy-ish.
I will get back to you with my answers your questions, but the first thing that struck me from your description was: Has someone else been fiddling with the amp circuitry before you? If the answer is "yes" or "probably" then you will do well to first have a long and thorough check of all components and solder joints.
Cheers,
Per
Last edited:
Here ya go:
As you noted, 2 output devices, 31V rails, no input OP amp, very high gain and the 0.22 ohm R647 at the output is not in the feedback path.
Also, no Miller cap C611 for Q611 (or Q607 on your amp).
Although there are 2 output pairs per channel, the back says 8 Ohm speakers (while the manual allows 4 ohm).
Not shown are the phono stage, the power supply (dual EI cores), the switches at the input and the second channel, of course.
Thanks, everyone. I appreciate the input
edit: I did check all caps and resistors. I'm not quite shure how to test the transistors, but both channels give exactly the same measurements. I think someone else may have fiddled with it before - the power cord is grey (which looks weird on a black amp) and has 3 instead of 2 leads (one now beeing connected to the chassis). I couldn't see this safety feature (and cause some of grounding-problems) in diagram...
As you noted, 2 output devices, 31V rails, no input OP amp, very high gain and the 0.22 ohm R647 at the output is not in the feedback path.
Also, no Miller cap C611 for Q611 (or Q607 on your amp).
Although there are 2 output pairs per channel, the back says 8 Ohm speakers (while the manual allows 4 ohm).
Not shown are the phono stage, the power supply (dual EI cores), the switches at the input and the second channel, of course.
Thanks, everyone. I appreciate the input
edit: I did check all caps and resistors. I'm not quite shure how to test the transistors, but both channels give exactly the same measurements. I think someone else may have fiddled with it before - the power cord is grey (which looks weird on a black amp) and has 3 instead of 2 leads (one now beeing connected to the chassis). I couldn't see this safety feature (and cause some of grounding-problems) in diagram...
Last edited:
Hi 1210,
Please understand that my comments are intended to help you. I have had to repair countless audio components that died during a "modification" attempt. Surely this costs more than doing things the right way. All I ask is that you do your best to do things properly. If there is a "not-to-expensive" tester you need to get or make, either wait till you can get it, or maybe someone will help you during the Christmas holidays acquire something.
Keep in mind that we have all been that starving student, myself included. Don't allow excitement and desire to rush you. You have all the time in the world on your side.
http://www.diyaudio.com/forums/solid-state/202047-roenders-fc-100-prototype-builders-thread.html#post2815121
The rest I'll leave you alone with. I suspect that you have the RC time constant wrong with the input filter as I don't think they would have affected signals in the normally accepted frequency range for human. It would not only sound dull, but the test specs would kill them in the marketplace. I just can't see any manufacturer doing that.
-Chris
Please understand that my comments are intended to help you. I have had to repair countless audio components that died during a "modification" attempt. Surely this costs more than doing things the right way. All I ask is that you do your best to do things properly. If there is a "not-to-expensive" tester you need to get or make, either wait till you can get it, or maybe someone will help you during the Christmas holidays acquire something.
Keep in mind that we have all been that starving student, myself included. Don't allow excitement and desire to rush you. You have all the time in the world on your side.
Thank you for that.
You should be able to get some okay matches with 10 ~ 20 transistors, and they are not expensive parts. The surplus will be used in experiments later on, so no waste. The test jig consists of a differential pair. It requires 4 resistors at the transistors to measure match. With reduced performance you can omit the current source and use a resistor for the tail current. The current source is not expensive to build, but I'm trying to keep it super simple for you. Here is a link that describes what I built. I find that I use it constantly.
http://www.diyaudio.com/forums/solid-state/202047-roenders-fc-100-prototype-builders-thread.html#post2815121
The rest I'll leave you alone with. I suspect that you have the RC time constant wrong with the input filter as I don't think they would have affected signals in the normally accepted frequency range for human. It would not only sound dull, but the test specs would kill them in the marketplace. I just can't see any manufacturer doing that.
-Chris
Hi Chris
Just out of curiosity: how many bias-setting multi-turn pots have you had to replace? The internet seems full of people doing exactly what I did - replacing single-turn "rherostats"/"beefy" pots with multi-turn pots, so shurely there must've been countless instances of failures on your desk...or shall I measure and report back about the currents involved, here?
I can't seem to find a supplier for them...other than ebay (where matched pairs cost like .50cents more/part).
As I said, I did look into matching myself. It's certainly not the complexity of the test-jig, but the difficulty to aquire NOS parts (at a reasonable cost). There are other, much more important and usefull things I'd like to get before I aquire a plethora of parts to use in later projects: a scope and a LC-meter. Maybe even a non-crappy multimeter (ideally with a built-in transistor-tester).
I didn't, either. I do now. Have a look at the input on the schematic I posted. I think it's part of that Rotel-sound.
Test specs? You mean in those magazines where the power and speaker cables change the sound very audibly and the manufacturers whose equipment is tested have ads in? Surely, the samples they get are "randomly" picked from the production line...
Sorry guys, due to work deadlines I just can't keep up with your thread speed.
I agree with much of what Chris has already said, hm … perhaps with a few exceptions,
So, if I may, I'll take things in stages, starting from the Line-in input to the Q601 input transistor circuitry.
Any reason not to use an el-cheapo/ebay ladder type 50K stepped attenuator here and just omit any balance function?
I never adjust the balance, unless there is a problem with the amp. And wait - your RA-820BX2 is probably the model with the notoriously ugly big shady gold/brownish combined volume/balance eyesore of a knob - right? I actually have such an amp on my repair to-do list and point Numero Uno is to replace it with a good quality standard 20k stereo Alps volume pot and a decently looking black knob, matching the rest of the otherwise nice 820BX2 amp design. Sod the “balance” feature.
And yes, you can't do without the DC blocks C601/602 as you'll never know what DC level could be connected to the line-in's. I also agree with your reduction in the “RF quench” Q601/2 base to ground capacitors from 470p. It has little to do with amp stability - I would have put it down to 100p or less.
See, if you listen at very low volume pot settings, the input signal is actually going through a, say, 19 (?) kOhm resistance loaded by the 470pF. That is a -3dB cut off at 17kHz! Ok, I know that most of us can't hear any low level sounds at this frequency anyway, but it is not Hi-Fi.
If you don't believe me, try to input a square wave at this low volume level and watch the amp output on an oscilloscope trace and you'll see what I mean.
Now, if you still get any “Radio Moskau” RF reception, try put in a 100pF to ground before the volume pot.
Next: To your other questions ....maybe tomorrow.
Per
I agree with much of what Chris has already said, hm … perhaps with a few exceptions,
So, if I may, I'll take things in stages, starting from the Line-in input to the Q601 input transistor circuitry.
Any reason not to use an el-cheapo/ebay ladder type 50K stepped attenuator here and just omit any balance function?
I never adjust the balance, unless there is a problem with the amp. And wait - your RA-820BX2 is probably the model with the notoriously ugly big shady gold/brownish combined volume/balance eyesore of a knob - right? I actually have such an amp on my repair to-do list and point Numero Uno is to replace it with a good quality standard 20k stereo Alps volume pot and a decently looking black knob, matching the rest of the otherwise nice 820BX2 amp design. Sod the “balance” feature.
And yes, you can't do without the DC blocks C601/602 as you'll never know what DC level could be connected to the line-in's. I also agree with your reduction in the “RF quench” Q601/2 base to ground capacitors from 470p. It has little to do with amp stability - I would have put it down to 100p or less.
See, if you listen at very low volume pot settings, the input signal is actually going through a, say, 19 (?) kOhm resistance loaded by the 470pF. That is a -3dB cut off at 17kHz! Ok, I know that most of us can't hear any low level sounds at this frequency anyway, but it is not Hi-Fi.
If you don't believe me, try to input a square wave at this low volume level and watch the amp output on an oscilloscope trace and you'll see what I mean.
Now, if you still get any “Radio Moskau” RF reception, try put in a 100pF to ground before the volume pot.
Next: To your other questions ....maybe tomorrow.
Per
Last edited:
Hi 1210,
What does the differential pair do at the beginning of the amplifier? It is the point where the input signal is compared to the output signal and the difference is subtracted to become the correction signal. It is critically important that this comparison takes place with a balanced, identical pair of transistors. If the match is imperfect, then the process of cancelling the distortion will not occur properly. This is the one place where you have a chance to affect how well this distortion reduction takes place. Other points in an amplifier can also affect distortion as the diff pair has limits as far as how well it can reduce distortion. So a match with outputs and drivers will also have a positive effect on the sound of the amplifier. Modern output transistors are far less variable in their characteristics then they used to be, so just by using good current parts, you can get a pretty close match in characteristics. Better if you match by hand, but still far better than the original output transistors. It's all up to you, your hands and your brain.
I have worked as an instrumentation technician and have replaced a ton of old multi-turn controls with same as a normal duty. We're talking about the expensive ones now. They become oxidized as much as open controls do. Multi-turn controls are always used as a potentiometer, a next to no current voltage divider. Bias circuits are not that, they do run current through the wiper which is why the wiper contact is so heavy (compared to a 10 turn control). Two different jobs for two different types of part.
-Chris
Nope. They come from the distributor in the country of the magazine. They know where the unit is going and it isn't uncommon for the piece to at least be tested. Sometimes the adjustments are touched up too. That makes a huge difference.
Then use something else that you can get new. Check Fairchild and On Semi for their versions.
Yeah, right! It would be rare to get decent matched transistors from a vendor who wants to move product only. In this world you have to fend for yourself. Check your parts and if matched - check the match.
No, it isn't difficult to match them. What this teaches you is knowledge on how things work. The entire exercise of improving a product is to do the detailed hand work. So this is a poor step to skip because the rewards are so high.
What does the differential pair do at the beginning of the amplifier? It is the point where the input signal is compared to the output signal and the difference is subtracted to become the correction signal. It is critically important that this comparison takes place with a balanced, identical pair of transistors. If the match is imperfect, then the process of cancelling the distortion will not occur properly. This is the one place where you have a chance to affect how well this distortion reduction takes place. Other points in an amplifier can also affect distortion as the diff pair has limits as far as how well it can reduce distortion. So a match with outputs and drivers will also have a positive effect on the sound of the amplifier. Modern output transistors are far less variable in their characteristics then they used to be, so just by using good current parts, you can get a pretty close match in characteristics. Better if you match by hand, but still far better than the original output transistors. It's all up to you, your hands and your brain.
Every one I come across in an "improved" piece of equipment. Some times the current through the wiper is substantial. You can figure the amount of current out from the schematic. I don't really have to since I have pulled a number of 10 turn pots out with open wiper contacts. Just think about this. In a multi-turn control, the geometry forces the parts to be smaller. The contact area is far less than a normal trim pot. That translates into far lower current limits. The other issue you brought up is curious to me. I have set bias currents for nearly 4 decades now. Some are touchy (Yamaha for example), while most are pretty easy. I can't understand there being a problem that requires much finer control than a normal single turn trimmer provides. If you really need finer control, increase the resistances of the resistors around the main terminals of the control and use a lower resistance trim pot. This is stone simple to do. Above all, use your head. Just because there is an idea currently that single turn controls are cheap (no good) doesn't make it so. Engineers select these parts and in production quantities a 10 turn trimmer is fairly inexpensive. They would use them if there was a problem with the single turn controls. However, those single turn controls are high quality parts, and they can be pricey. Use the proper part for the job and avoid trouble down the road.
Every amplifier is different, as is every control. If you are using Ebay controls from China, expect a much lower current rating for the moving contact. The wiper current should be something you can calculate if you wish. So don't try and use the specs from a good control and apply those to the ones you buy inexpensively. Attempt to get the specs for the parts you are actually using. This is not a debate to be won cleverly, it is engineering and common sense.
I have worked as an instrumentation technician and have replaced a ton of old multi-turn controls with same as a normal duty. We're talking about the expensive ones now. They become oxidized as much as open controls do. Multi-turn controls are always used as a potentiometer, a next to no current voltage divider. Bias circuits are not that, they do run current through the wiper which is why the wiper contact is so heavy (compared to a 10 turn control). Two different jobs for two different types of part.
-Chris