Hi AngelP,
I would never leave the speakers unprotected in the long run. Currently I have an outboard dc protection relay. I wouldn't be selling these amps anytime soon. Probably will end up in my will !
I know! I liked the amp ever since I got it, there was something I could not figure out why the amp sounded a certain way. even though I tried to fix it by replacing caps and diodes in the power supply, adjusting bias, dc offset, replaced signal capacitors at several places and nothing fixed it, it improved, but still, I even disabled completely the preamp/ tone board and just bypassed it. but removing that resistor and adding more filtering caps has pretty much solved it, it was like removing a fine blanket on top of my speakers. Crazy! it's great! that's why I am curious to see if the current mirror and VAS upgrade will take it to another level, although, it is good the way it is now.
I have not checked how well those trannies match, but I plan on doing it this weekend and of course, if they are off, I will replace them, but the question is how much off do they need to be to need replacement, do they need to be a perfect 100% match?
I think either way, I will order new trannies and yes I will replace them to see what changes I get, if any. Hey, this is our hobby and is the right thing to do!
Last edited:
wow, i have a hard time beleiveing that the 47k value of R613 shown on the schematic is not a mistake.
did you happen to measure the value of the actual resistor that was removed before disposing of it?
mlloyd1
did you happen to measure the value of the actual resistor that was removed before disposing of it?
mlloyd1
Honestly, mlloyd1 - why would anyone put a 47k in that position??? Or even a single 47E for that sake?
If you have a plausible suggestion, I would be most happy to hear about it and learn.
Me, I suspect that there were things lost in translation from Curtis' schematic UK drafts to the guys in Taiwan.
tell me about it! I was not expecting that, in all honesty, I was expecting a 47k reading on my meter,
so I kept reading the actual value, once removed, over and over again.
I did not dispose of it as it is good.
I have not checked how well those trannies match, but I plan on doing it this weekend and of course, if they are off, I will replace them, but the question is how much off do they need to be to need replacement, do they need to be a perfect 100% match?
I think either way, I will order new trannies and yes I will replace them to see what changes I get, if any. Hey, this is our hobby and is the right thing to do!
I have the luxury of having a reel of 2SA992F (albeit pretty thinned out by now), so I normally match to about 0.5%.
Some consider a 1% match to be ok (e.g. hFE 402 vs. 398).
The tricky part is to maintain the transistors at the exact same temperature when measuring - just pinching the TO-92 between your fingers will increase its hFE by 2-3%.
This is also why it is a good idea to thermally connect the input pair on the pcb, be it by superglue or heat shrink tube (see pic in 216).
Re the "47k" resistor: Had that been the true value it would have caused a big voltage drop across it and the amp output could have been driven into a "speaker killer" DC level.
The 47E resistor may have been an attempt to reduce the Rotel's DC offset at the speaker terminals - I don't know, just guessing - even so, that is not a good way of doing it. Balanced input pairs should be exactly that - balanced.
Can the difference be heard on pair matched well vs a close enough matched pair? apparently there is jig to do this matching that other people mention here, I will search about it. all I use is my DMM which it has an hFE test setup I use.
so I pulled out the 2 pairs on my rotel and these are the readings, left pair 245 and 248, right pair 233 and 246.
then I matched some KSC1845's and selected the ones that matched the best on what I had and used those. I noticed no difference on a quick listen.
so I pulled out the 2 pairs on my rotel and these are the readings, left pair 245 and 248, right pair 233 and 246.
then I matched some KSC1845's and selected the ones that matched the best on what I had and used those. I noticed no difference on a quick listen.
That is actually not bad at all - definitely not for the left pair.
I have seen far worse on Rotel amps.
Your DMM setup is fine, it is not the actual value that is important, just the match. No need for a jig, just snug the pair into a small piece of tubing, face to face to keep them at the same temperature and re-measure each device twice.
Audible? Well, a good match will definitely reduce THD and you are now so deep into the amp, so it would be silly not to do the job.
Oh, and surely you are not putting KSC1845 NPN's in for the 2SA1016 PNP pair?
Silly me! You are correct, I meant to say KSA992!
I was also wondering the feedback resistor on your Rotel RA-820AX R637 8.2K, is that the original value? mine has 39k in that position.
I have been playing around replacing that resistors with different values and using precision resistors does make a difference... I am having fun with this stuff man!
I was also wondering the feedback resistor on your Rotel RA-820AX R637 8.2K, is that the original value? mine has 39k in that position.
I have been playing around replacing that resistors with different values and using precision resistors does make a difference... I am having fun with this stuff man!
Actually, if you look closely your feedback resistor circuit is 12k in parallel with 39k, which gives somewhere about 9k - as long as the fuse is intact.
Yes, all this soldering, learning and modding is definitely great fun.
However, the real fulfilment comes with the peaceful intense satisfaction when you listen to what your efforts have actually achieved.
And even more so when others suddenly can hear the difference too - the transformation from a tech geek to an audio hero.
Yes, all this soldering, learning and modding is definitely great fun.
However, the real fulfilment comes with the peaceful intense satisfaction when you listen to what your efforts have actually achieved.
And even more so when others suddenly can hear the difference too - the transformation from a tech geek to an audio hero.
Hi Per,
The jig is the only way to really get a close match. Difficult with smd parts as you need to make an adaptor to hold the pair down on the contacts.
I know you're doing things as carefully as possible, but I was in the same boat years ago and was quite disheartened when I retested my early "matches". It did explain the hit and miss results I was getting. Just build the jig and be sure about it. In addition, with the jig, you can run the tail current close to the actual conditions of the target circuit. It's adjustable.
With pairs that close you need to match any degeneration resistors if used or they will throw the match off.
The diff pair(s) are critical where performance is concerned. This is where the signal is compared against the input signal and distortion components are extracted and cancelled. If this isn't balanced, then your output will contain more distortion than it should. Low DC offset (assuming the circuit is properly designed) is a happy byproduct of a closely matched differential pair. Beware that using a servo to create low DC offset will tend to unbalance your diff pair if implemented in the usual way.
-Chris
The jig is the only way to really get a close match. Difficult with smd parts as you need to make an adaptor to hold the pair down on the contacts.
I know you're doing things as carefully as possible, but I was in the same boat years ago and was quite disheartened when I retested my early "matches". It did explain the hit and miss results I was getting. Just build the jig and be sure about it. In addition, with the jig, you can run the tail current close to the actual conditions of the target circuit. It's adjustable.
With pairs that close you need to match any degeneration resistors if used or they will throw the match off.
The diff pair(s) are critical where performance is concerned. This is where the signal is compared against the input signal and distortion components are extracted and cancelled. If this isn't balanced, then your output will contain more distortion than it should. Low DC offset (assuming the circuit is properly designed) is a happy byproduct of a closely matched differential pair. Beware that using a servo to create low DC offset will tend to unbalance your diff pair if implemented in the usual way.
-Chris
Hi Chris,
Yes, but I thought that it may be a bit much work if you're only modding one amp. It can be done with a piece of insulated tubing, but ok, it does take time, care and patience.
On the subject, have you ever re-measured a known well matched pair after, say, a couple of months of power-on? The hFE will probably settle in, but if you make sure to pick pairs from the same reel (production batch) they should (?) maintain their match?
Now, being the owner of a small SMD oven, I was planning to try to anneal a matched pair at 100 degC for 12h - or similar - to check out whether there is any initial drift difference. Any thoughts on this?
Per
Yes, but I thought that it may be a bit much work if you're only modding one amp. It can be done with a piece of insulated tubing, but ok, it does take time, care and patience.
On the subject, have you ever re-measured a known well matched pair after, say, a couple of months of power-on? The hFE will probably settle in, but if you make sure to pick pairs from the same reel (production batch) they should (?) maintain their match?
Now, being the owner of a small SMD oven, I was planning to try to anneal a matched pair at 100 degC for 12h - or similar - to check out whether there is any initial drift difference. Any thoughts on this?
Per
Hi Per,
Yes, I have measured previously matched transistors with the jig in two different situations. One where they just sat in storage (no change in match) and others that were in service for a few years. No change in match there either, however if that amplifier goes DC, then there are big changes in the match. I'm pretty sure the noise performance of the one that was reverse biased would also be degraded along with the lower beta.
It seems like we are both curious people!
As for one project and the matcher being too much work, well. What can I say? You get out of it what you put in. The first one I made was for a project and I had recently not been happy with the measured beta and the resulting performance. So I made just the NPN section on perf board with fixed tail current (but still a CCS). Testing the other matches I made, only a few were even close to being acceptable. So I rematched those and a few more. The results with actual amplifiers and measurements convinced me pretty solidly that doing it this way was really the only way if it matters. If you are building your own amp, it does matter (or why bother?). Later I had to do a PNP pair, so I made the NPN and PNP version with selectable tail current. I used it so much that I ended up designing a PCB for it, then a version using DIP switches. Other members here have designed their own PCB, one was Cogeniac who was kind enough to send me a board to build and check it out. His board works great, and he made it larger than mine so it is easier to handle.
About the most expensive parts were the 0.1% resistors I use for collector (drain) loads and the base (gate) connections to the ground. That gives me the ability to measure the currents for a BJT to determine the actual, real beta - just to satisfy my curiosity. The 0.1% resistors also allow for very, very tight matches (because I further matched those for even tighter pairs using an HP 34401A. It doesn't have that accuracy, but you can compare pairs of resistors very accurately. These days I use a 3457A for this work, both in a Kelvin configuration of course!
If a member bought a bunch of 1% resistors and matched those, the matches would be very close too. So you can see, it isn't really much effort to make a quick little jig to handle one polarity of transistor at a fixed current. Just order those parts when you order the stuff for your work on the amp.
-Chris
Yes, I have measured previously matched transistors with the jig in two different situations. One where they just sat in storage (no change in match) and others that were in service for a few years. No change in match there either, however if that amplifier goes DC, then there are big changes in the match. I'm pretty sure the noise performance of the one that was reverse biased would also be degraded along with the lower beta.
It seems like we are both curious people!
As for one project and the matcher being too much work, well. What can I say? You get out of it what you put in. The first one I made was for a project and I had recently not been happy with the measured beta and the resulting performance. So I made just the NPN section on perf board with fixed tail current (but still a CCS). Testing the other matches I made, only a few were even close to being acceptable. So I rematched those and a few more. The results with actual amplifiers and measurements convinced me pretty solidly that doing it this way was really the only way if it matters. If you are building your own amp, it does matter (or why bother?). Later I had to do a PNP pair, so I made the NPN and PNP version with selectable tail current. I used it so much that I ended up designing a PCB for it, then a version using DIP switches. Other members here have designed their own PCB, one was Cogeniac who was kind enough to send me a board to build and check it out. His board works great, and he made it larger than mine so it is easier to handle.
About the most expensive parts were the 0.1% resistors I use for collector (drain) loads and the base (gate) connections to the ground. That gives me the ability to measure the currents for a BJT to determine the actual, real beta - just to satisfy my curiosity. The 0.1% resistors also allow for very, very tight matches (because I further matched those for even tighter pairs using an HP 34401A. It doesn't have that accuracy, but you can compare pairs of resistors very accurately. These days I use a 3457A for this work, both in a Kelvin configuration of course!
If a member bought a bunch of 1% resistors and matched those, the matches would be very close too. So you can see, it isn't really much effort to make a quick little jig to handle one polarity of transistor at a fixed current. Just order those parts when you order the stuff for your work on the amp.
-Chris
Hi again memofer,
Just catching up on how you're getting along with the mods.
Your experience above of "a completely different amp" has been echoed by every single one of the Rotel owners I now have done the mods for.
If you feel adventurous and willing to report back on the thread on your results, I now have a spare set of prototype current mirrors and VAS transistor replacement pcb's for you to try out for free. I will of course provide detailed instructions on how to mount them in your amp.
Your call.
Hi Angelp,
The amp is doing well and the mods are working great, no issues at all. Actually, I added DC offset pots to dial a more accurate offset and I can get it pretty close to +- 1mV.
Nice to know your customers/friends are happy with your mods, says a lot about you! mods worked great!
That's great, yes, I am willing to try your prototype, send them my way,
appreciated!
The amp is doing well and the mods are working great, no issues at all. Actually, I added DC offset pots to dial a more accurate offset and I can get it pretty close to +- 1mV.
Nice to know your customers/friends are happy with your mods, says a lot about you! mods worked great!
That's great, yes, I am willing to try your prototype, send them my way,
appreciated!
Last edited:
I was also wondering, do you think it would be worth it to add a separate, regulated, (a lot cleaner) power supply (42 volts) to the differential pair and VAS so to separate it from they main supply, after all the main supply for the output transistors is not that clean?
I hear from others here that it makes a big difference and I am curious, but lazy to apply it, but I think that could be another nice project to tackle in the future, have you done it?, if so, did you like results? thanks,
I hear from others here that it makes a big difference and I am curious, but lazy to apply it, but I think that could be another nice project to tackle in the future, have you done it?, if so, did you like results? thanks,
Last edited:
Yes, any cleaner power supply for the input stages would of course improve things. I did try to power the RA-820 amp from a separate, improved RA-931's PS on the bench - and it did make a difference.
But IMHO building a separate regulated supply box for your project would be over the top. The simple RC filters which I put in reduced the input stage ac ripple from about 30mV to 3mV, and the better balancing of the input stage does further improved the PSRR to a point where furher exotic "improvements" become pretty much marginal.
Just my 2p.
But IMHO building a separate regulated supply box for your project would be over the top. The simple RC filters which I put in reduced the input stage ac ripple from about 30mV to 3mV, and the better balancing of the input stage does further improved the PSRR to a point where furher exotic "improvements" become pretty much marginal.
Just my 2p.
Hi Per,
Yes, separate supplies for the front end can make a huge difference to the positive! If you had a little extra voltage, you could regulate the front end supplies so they don't ever see bouncing voltage. If the front end remains powered up for longer, that should reduce or completely eliminate turn-off thumps.
-Chris
Yes, separate supplies for the front end can make a huge difference to the positive! If you had a little extra voltage, you could regulate the front end supplies so they don't ever see bouncing voltage. If the front end remains powered up for longer, that should reduce or completely eliminate turn-off thumps.
-Chris
In the long run, I think I am going to look into regulating the front end without adding an external supply. I have another vintage receiver, that I could use as a guinea pig.
what would be the best approach though? regulator IC or transistors as emitter follower and such..
then again, I am very happy the way it sounds right now, I may never get there...
what would be the best approach though? regulator IC or transistors as emitter follower and such..
then again, I am very happy the way it sounds right now, I may never get there...
Last edited: