Improve a Rotel amp THD by 20dB!

Member
Joined 2012
Paid Member
Thanks Per, I trust your judgement. I'm thinking of having a play with the Opamps mentioned and adjusting the Zeners to suit. I have those Opamps available and sockets.

I have about 50 amps in the storage unit that I'm slowly restoring, if needed, it's not the most ideal humidity. I've noticed on some CA and NAD gear from the 90s has bad corrosion so far. This wasn't as bad tbf. Interesting about Sony's, I have some Sony gear there.

I'll check out the Dowsil 3140, not heard of that before and seems you have a good process there.

This amp has the Black Gates caps. I wanted to see what the fuss is about them. I had replaced them with Nichicon Muse and Fine Gold's I had spare. I couldn't tell much difference. I tried them in another amp though, which had new caps, and made a nice change to the lower frequencies upto about 1kHz. More dynamic. I have a Parasound Pass Labs DAC with lots of Black Gate caps that I was thinking of replacing for new but I think I'll leave that now. I like that DAC alot, as with my other modern DAC's.

I'll take a drive to Peak this week to get a new Atlas ESR metre, it's not far from me. Do you recommend any of their other equipment they sell?

50 amps in storage !!! I will promptly relay this to the Supreme Command, maybe she will look at my queue of 8 Rotels waiting for upgrades with milder eyes.:)

You are absolutely right about old CA and NADs corroding - for some reason they employed the thinnest Ni coating. I will try to find a pic of the internal pcb of the old SONYs, but enclose a few of the back panel of a NAD3030 - before and after the repair.

I find that all of Peak devices work extremely well, and their after sales service is great too. I think it was my DCA55 or ESR60 that developed some problem - and they fixed it and updated the software version on top of that.

Per
 

Attachments

  • DSCF4495.JPG
    DSCF4495.JPG
    890.5 KB · Views: 183
  • DSCF4568.JPG
    DSCF4568.JPG
    892.2 KB · Views: 190
Member
Joined 2012
Paid Member
1. And most important: Very interesting thread!
Thank you very much.:up:

2. Some time ago you published a composite VAS including a third T3. In your example, T3 acts at the same time as the buffer for T2 and also as a "current splitter" by T3's VBE paralleled by the 330 ohm resistor. In your example, this might set the IC for T2 as some 0.7/330=2.1mA. I first saw this option in D.Self's power amplifier book, and then tried it in a Yamaha AX590 VAS modification. But that lead to some oscillation, so I discarded it again. My first question then: Do you still use this theoretically elegant modification, e.g. is it successfully long-term stable/safe in your Rotel mod's?
I am very surprised that you got any oscillation from adding an emitter follower in the circuit. I have now upgraded umpteen Rotels with the VAS3 module without any issues - and I have indeed tested and subjected the amp to some of the most difficult loading combinations I could think of.

Maybe think of it like this - if the amps has two driver stages (as eg. the RA-971) - the first low power driver stage acts as an emitter follower to reduce the load on the VAS to uA levels. You wouldn't expect any problems from this?
I did however have oscillation issues when I simply tried to replace the original 2SC1941 VAS transistor with a darlington type (to do a quick EF-VAS fix). Probably because of feedback into the EF from the collector not tied to a stable V+ or GND.


3. Did you try to include improved CM's in your modifications? I am especially interested in the EFA current mirror as published in Samuel Groner's 2011 comments on Self's designs. I have not tried this one out, and I am wondering why I am not able to find any implemented and working examples of such avariant when searching by Google. Does this EFA CM variant come along with serious drawbacks in practice?
On the RA-931 the simple CM1 current mirror works a treat.

I am in the process of trying out an EFA version on a RA-971 which due to its double differential LTP and dual VAS design needs it and will post here on my findings as and when.

4. Did you try to include a output-stage including Miller Cdom variant? I did this in a Yamaha AX592 and it works fine. In this mod, the Cdom of 100pF is split into two serial 220pF, and the midpoint is connected to the power device's emitters resistors's common connection point by a 1k4 resistor. Result: See the two Arta graphs.
Yes, and also found the improvement to be marginal.

5. By which criteria did you choose the different emitter degeneration's resistors values of the IPS, CM and the VAS? E.g. is there a "sweet spot" to do so? Depending on the Ic's of the IPS, the CM and the VAS I arbitrarily chose to set them for a compromise 1:10 ratio of re/R_degen. I am shure there must be better criteria than this rather irrational magic of decimal values ...
The main thing is to get the DC voltages right so the EF-VAS actually turns on properly - but you have many ways to do that, vary the emitter resistors on the CM or increase the LTP current. There may be a "sweet spot", but I rather think of it as a "sweet range" with very similar improvement values.

And I kept the original Rotel component choices as far as possible - a) because Curtis is a proven, excellent designer and b) I wanted to keep the upgrade as simple as possible so most diyAudio people could do it and c) because I am a lazy bu**er at heart:D

I hope not to bother you too much with these questions!
That is the main reason I am writing on this thread. Hoping that I can share some knowledge - but definitely to learn from other posters questions and experience.

Per
 
50 amps in storage !!! I will promptly relay this to the Supreme Command, maybe she will look at my queue of 8 Rotels waiting for upgrades with milder eyes.:)

You are absolutely right about old CA and NADs corroding - for some reason they employed the thinnest Ni coating. I will try to find a pic of the internal pcb of the old SONYs, but enclose a few of the back panel of a NAD3030 - before and after the repair.

I find that all of Peak devices work extremely well, and their after sales service is great too. I think it was my DCA55 or ESR60 that developed some problem - and they fixed it and updated the software version on top of that.

Per

Thanks, Per

You did really well cleaning up those RCAs. I like the use a drill with a pen lid losely stuffed with some ultrafine wire wool and metal polish inside in place of a drill bit. Seems strange yet does the job rapidly and best results I've had.

I can't remember which Sony models I have. I'm not the best with remembering models, there are that many. I know I have two ES series. One from about '80 and a 90s unit.

Looking at my RA-930BX after reading the posts here you mentioned...

I have no Opamp preamp, just phono stage and tone controls.

C601/602 10uF electrolytic you replaced with wire links, am I ok doing this?
Also, I like to replace emitter resistors as I seem to get slightly better balance when matching closely and going up in wattage. These are cement types, are non-inductive wirewound worthy?
 
Member
Joined 2012
Paid Member
Hi,
Heavens no, the old RCA's in the NAD were way past their use-by date and pitted beyond polishing up. And if you try to pull them out like rotten teeth, their brittle plastic base will crack.

So I designed and drilled an aluminium panel, painted it satin black and put new gold plated RCA's in. If more NAD3030's come through my workbench I will consider to have a double sided FR4 board made which would make things a lot easier.




Ah, that explains things regarding the 930BX, I have often wondered what the difference was to the 930AX, as the tech manual for the BX seems to be very hard to find. Would you by any chance have a copy of the TM?
If there is no preamp stage, then the entire voltage amplification is in the power stage and you'll need a capacitor at the input to safeguard against possible input signal DC. So no, I would not recommend removing C601/2.

I don't think that you can find many non-inductive wirewounds, the original Rotel green 0.22 ohm resistors I believe are metal film - or you could put in metal plate KOAs (other makes are available), these are both non-inductive and work well.

Per
 
Hi Per,
just curious, have you tried to disconnect driver's emitters from the output stage emitters...
Drivers and outputs emitters connected together are no longer used for a long time and it might be for a reason.

Disconnecting would lower "step" distortion by digit or couple, thought might be interesting to try how it measures.

I have tried to hear difference in changing that connection, - no hearable difference, but might be visible on numbers...

Kind regards
Kes
 

Attachments

  • 2021-03-30 18_57_27-hfe_rotel_ra-921_service.pdf (SECURED) - Adobe Acrobat Reader DC (32-bit).jpg
    2021-03-30 18_57_27-hfe_rotel_ra-921_service.pdf (SECURED) - Adobe Acrobat Reader DC (32-bit).jpg
    162.5 KB · Views: 225
Thanks again Per. BTW your wife comments are hilarious, I'm guessing she doesn't do as many posts on here as you LOL

That's exactly what I was thinking about the DC on the amp's input, I'm glad I asked. I haven't got a schematic, unfortunately. I briefly looked for one and if I find one I'll post it.

Apart from the preamp Opamp circuit, it appears the same, from what I can tell so far. I'll try to upload some photos when I figure this temporary phone out (not the best camera quality)

Am I ok replacing the phono stage output caps (miniature yellow Elna's 10uF) to 1uF Wima's though aren't I, or did you do that mod because you replaced the Opamp, which I'm not going to do.
 
Member
Joined 2012
Paid Member
Hi Per,
just curious, have you tried to disconnect driver's emitters from the output stage emitters...
Drivers and outputs emitters connected together are no longer used for a long time and it might be for a reason.

Disconnecting would lower "step" distortion by digit or couple, thought might be interesting to try how it measures.

I have tried to hear difference in changing that connection, - no hearable difference, but might be visible on numbers...

Kind regards
Kes

Hi Kes,
I think that is a difference between two 'schools' of power stage thinking, some like it floating and some prefer a wee bit of ground anchorage.
If the amp is well balanced it really doesn't matter which. And I wouldn't expect any sound difference.

Per
 
Hi Per

...
I am very surprised that you got any oscillation from adding an emitter follower in the circuit
...
I did however have oscillation issues when I simply tried to replace the original 2SC1941 VAS transistor with a darlington type (to do a quick EF-VAS fix). Probably because of feedback into the EF from the collector not tied to a stable V+ or GND.
...
Your latter observation might match my oscillation's experience. Maybe this was a local problem inside the 3-transistors-setup. From today's point of view unfortunately I did not try to track down it's cause by further tweaking the T3 option. Meanwhile, I saw that you refined the C's in a later iteration, adding 27pF between EF's base and VAS's Collector. This is now the way I will follow. I would have much appreciated if D.Self, who does mention this buffering variation in his great book, had further empirically examined it's practical behavior/implementation. Like he did (much appreciated by my!) for so many other details.

...
I am in the process of trying out an EFA version on a RA-971 which due to its double differential LTP and dual VAS design needs it and will post here on my findings as and when.
...
Referring once again to his book, D.Self has published some Tables and Graphs about the consequences of current symmetry mismatch for the input pair. You certainly are aware of them. It seems impressive to me that even a slight current mismatch of 0.5% does impair the accuracy of this stage. Therefor I am quite eager to hear from your further iterations with the CM. My feeling is that it is quite worth to polish the CM's accuracy as much and as practical as possible. Even if there may be only slight consequences for the overall THD numbers.

...
And I kept the original Rotel component choices as far as possible - a) because Curtis is a proven, excellent designer and b) I wanted to keep the upgrade as simple as possible so most diyAudio people could do it
...

This is a very respectful, honorable and community-friendly decision/approach. And also an esthetically clean, simple, minimal-art-like one! A nice example of best engineering practice: get the most out of minimum cost. Cudos!

My mentionned AX592 mod instead looks quite messy, and is still non-linear "work under progress" and therefore a steady construction aera. Maybe even featuring an exponent of "the good, the bad and the ugly" - take your choice when looking at the picture ... And also therefore your inputs here are greatly appreciated.

Simon
 

Attachments

  • WUP.jpg
    WUP.jpg
    209.8 KB · Views: 202
Hi Simon,
just as coincidence I have 2 Yamahas at home so I can compare sound of them and I can tell that earlier model AX-400 sounds more interestingly (for my taste) than AX-450. AX-400 sounds a bit open and a bit wider.

One significant change between those series - Yamaha in newer series introduced "EF-VAS with inverted EF" while in AX-400 it was "Cascode VAS". And the same "Cascode VAS" was used in Marantz PM80mkII which I also have and can confirm that it sounds very good.
The other change AX-400 has a bootstrap, so maybe some amount of bootstrapped PFB gives to sound something interesting...

Per,
have you tried by any chance "cascode vas" in your VAS prototyping stage? Wonder what was the outcome compared to your final product...

I see couple of "Cascode VAS" benefits for DIY'ers:
1. minimum additional parts.
2. lower transistor opens possibility of rolling various low-voltage hi-hFe transistors. Even to try FETs.

As I have not tried "cascode vas" yet in Rotel, not sure if by introducing upper cascode transistor would not disbalance biasing circuit...

Kes
 

Attachments

  • 2021-03-31 11_31_15-Yamaha-AX-400-Service-Manual.pdf - Adobe Acrobat Reader DC (32-bit).jpg
    2021-03-31 11_31_15-Yamaha-AX-400-Service-Manual.pdf - Adobe Acrobat Reader DC (32-bit).jpg
    43.6 KB · Views: 172
Last edited:
Member
Joined 2012
Paid Member
Hi Per
Therefor I am quite eager to hear from your further iterations with the CM. My feeling is that it is quite worth to polish the CM's accuracy as much and as practical as possible. Even if there may be only slight consequences for the overall THD numbers.

Simon

Yes, I am afraid that the RA-971 is putting up a vicious fight to keep me from studying current mirrors in the input. I've already gone through quite a few blown fuses and transistors, so I will have to sit back (with a beer supply) to plan another angle of attack.

You may also take a read in Bob Cordell's great book (thank you Joe from Hamburg for pointing me to that), he describes the issue quite well although not in as much detail as Doug Self normally does.

But as far as I remember, even Self stopped short of a full analysis of the issue at hand with CM's in a double differential circuit.
I will definitely post the findings (good and bad) here as I find it an interesting challenge - which will keep me safely off the street corners for some time - although not for long, of course.;)

You know, I can still wolf whistle at passing ladies - I just can't remember why anymore.....:scratch:

Per
 
Member
Joined 2012
Paid Member
Hi Simon,
just as coincidence I have 2 Yamahas at home

Per,
have you tried by any chance "cascode vas" in your VAS prototyping stage? Wonder what was the outcome compared to your final product...

I see couple of "Cascode VAS" benefits for DIY'ers:
1. minimum additional parts.
2. lower transistor opens possibility of rolling various low-voltage hi-hFe transistors. Even to try FETs.

As I have not tried "cascode vas" yet in Rotel, not sure if by introducing upper cascode transistor would not disbalance biasing circuit...

Kes


Crikey! 50 amps and another two Yamaha's at home - how do you get around in the house?:rolleyes:

You may find that at the start of this thread I tried cascoding in the LTP current source. It did work extremely well, but I found it quite fiddly to do in practice.
For the same reason I haven't done any cascoding in the VAS stage.
And even with a hi-hFE VAS transistor there will still be loading of the input. My VAS3 module has an equivalent hFE of about 3000.

Per
 
...
You may also take a read in Bob Cordell's great book (thank you Joe from Hamburg for pointing me to that), he describes the issue quite well although not in as much detail as Doug Self normally does.
...

Yes. And S.Groner, in another publcation, refers to a Paper of D. P. Laude: A Low-Noise High-Precision Operational Amplifier, IEEEJ. Solid-State Circuits, vol. SC-16, no. 6, pp. 748–750, December 1981. Unfortunately I was not able to get hold of this one.

Samuel Groner's Paper "Comments on Audio Power Amplifier Design Handbook by Douglas Self" (2011) said:
...
Better cancellation and more freedom for VAS biasing is achieved with theEFA mirror (figure 1). If R5 is chosen such that Q1 and Q2 are operated atequal collector voltage and R3 and R4 have equal value the error introducedby the base current of Q5 is cancelled with the base current of Q6. Overallthis topology leads to very good input stage balance and offers low drift (seee.g. [1]).
...
This scheme to balance out the input stage currents seems very appealing to me. But, as neither me, nor you have tried this flawor (yet?), this discussion becomes rather more and more theoratical, now. And this thread here instead started and bewared itself as a refreshingly practical thread ... No practical experciences with this one, anyone?

Maybe it might be better to move these questions into anothter thread, more prone to theoretical design issues? Anyway - The attached picture is from Groner's paper, also, and refers to to the aforementionned text. I Hope this is ok from the copyright's point of view.
 

Attachments

  • EF_Groner.jpg
    EF_Groner.jpg
    29.4 KB · Views: 504
...
I tried cascoding in the LTP current source. It did work extremely well, but I found it quite fiddly to do in practice.
...

For the same reason I haven't done any cascoding in the VAS stage.
...
Could you explain please: What was fiddly with your cascoded LTP current source? Was it the physical setup? Or some values tweaking? Something else? My own iteration of black-boxed, cascoded LTP current sources with +Rail-RC-filtering look quite fiddly/ugly, indeed.

Then Cascode vs. EF for the VAS: I would distinctly vote in favor of the EF variant, because of the two and stacked EF-VAS VBE's giving the CM more VCE to breath than a single (cascoded of not) VAS VBE ... This better CM VCE headroom was also the reason why I modded the exotic looking PNP EF of my AX592 into a more familiar NPN EF: The original PNP-variant for the EF was even worse than a single (cascoded or not) VAS in terms of VCE headroom for the CM. And also this PNP-to-NPN mod, ressembling some kind of gender reassignment for the EF, was some kind of fiddly ... Life is not meant to be easy, indeed.
 

Attachments

  • VAS.jpg
    VAS.jpg
    101.9 KB · Views: 527
  • CS.jpg
    CS.jpg
    60.2 KB · Views: 527
Last edited:
Rotel RB-990-BX

Hi,

I have a Rotel RB-990-BX (200w/ch) that I normally use for driving my aux speakers which are KEF Reference Series Model Ones but I've been using it to drive my KEF Reference Series Model Threes while trying to get my Meridian 557 amp fixed. Although the Rotel doesn't have the clarity and presence of the Meridian, I am pleasantly surprised at how well it drives the threes. I wish it had balanced inputs though. I'm driving the amps with a Meridian 502 balanced preamp (both the 502 and 557 are balanced, dual mono).

I was wondering if anyone here has worked on a Meridian 556, 557 or 559 since I think they all have the same auto bias and protection circuit design. I purchased my 557 new in 1968 and it's worked perfectly until earlier this month when it refused to fully power on. It will go through the normal soft start sequence, sit there for about 2 seconds and then start the protection shut down sequence. It's definitely not the output stage and everything else looks good except for the bias voltage on the right channel is a bit high. Based on the age of the amp and the readings I see, I'm pretty sure it's one or more electrolytic caps in the auto-bias circuit or perhaps the large power caps off the relay board but I don't want to just start replacing things.

Has anyone seen these symptoms and have any suggestions for troubleshooting the problem with this amp?

Thanks,

Dean
 
Member
Joined 2012
Paid Member
Yes. And S.Groner, in another publcation, refers to a Paper of D. P. Laude: A Low-Noise High-Precision Operational Amplifier, IEEEJ. Solid-State Circuits, vol. SC-16, no. 6, pp. 748–750, December 1981. Unfortunately I was not able to get hold of this one.


This scheme to balance out the input stage currents seems very appealing to me. But, as neither me, nor you have tried this flawor (yet?), this discussion becomes rather more and more theoratical, now. And this thread here instead started and bewared itself as a refreshingly practical thread ... No practical experciences with this one, anyone?

Maybe it might be better to move these questions into anothter thread, more prone to theoretical design issues? Anyway - The attached picture is from Groner's paper, also, and refers to to the aforementionned text. I Hope this is ok from the copyright's point of view.

I can assure you that for me it is currently indeed a very practical issue, see pics.
Cordell suggests adding emitter followers on both the NPN and PNP current mirrors - but notes that due to component differences it could lead to some infighting over the current through the two VAS transistors.

He indicates that that would lead to an offset - for me it has led to blue smoke.:bawling:

So, my:idea: was to only use EFA in one of the mirrors - which should set a single defined VAS current.
And that is where I am at the moment.

I think someone (was it sgross?) mentioned that this issue had been covered in a thread here on diyAudio, but I haven't been able to find it.


And yes, I found the LTP cascode solution to be a practical issue, easier to do on paper than pcb.

Then Cascode vs. EF for the VAS: I would distinctly vote in favor of the EF variant, because of the two and stacked EF-VAS VBE's giving the CM more VCE to breath than a single (cascoded of not) VAS VBE ... This better CM VCE headroom was also the reason why I modded the exotic looking PNP EF of my AX592 into a more familiar NPN EF: The original PNP-variant for the EF was even worse than a single (cascoded or not) VAS in terms of VCE headroom for the CM. And also this PNP-to-NPN mod, ressembling some kind of gender reassignment for the EF, was some kind of fiddly ... Life is not meant to be easy, indeed.
He, he ...try reading that sentence out loud quickly 5 times after a drink or two.:p

Or add a bass and drum track - and we could get an instant rap music hit? (With lyrics that makes sense for a change!)

Sorry, no offense intended Simon, I know that I am also all to guilty of acronym juggling:joker:

Per
 

Attachments

  • DSCF4570.JPG
    DSCF4570.JPG
    920.6 KB · Views: 483
  • DSCF4569.JPG
    DSCF4569.JPG
    941.1 KB · Views: 466
Do you know these two links?

1. Featuring mirrors only:
Current Mirror Discussion
Much theoretical talk there, but no data of real-world implementations. Have a special look at Mark Johnson's thumbnail at post No. 327 (which is re-thumbnailed here)

2. Featuring mirrors and about VAS current setting by tweaking mirror(s) components:
2016 Burning Amp Festival speaker: Bob Cordell - YouTube
Bob Cordell in full action at the burning amp fest 2016, with focus on enhanced mirrors especially from 17:20, from 33:55 and from 47:00

After all this current reading and mirror watching and along with a modder's spirit, M.Johnson's thumbnail made me think of the EF (Q4 on M.Johnson's schema) as a component potentially nicely fitting next to the CM onto the same modding-boardie. Along with some more components providing an extended functionality. I had this idea for four main reasons:

1. Because both EF (Q4) and CM "helper" transistor (Q3) could be seen as a functional, mutually depending pair providing current and voltage symmetry for the input LTP's collectors. Therefore both Q3 and Q4 can be regarded as functional partners providing a further improved mirror functionality.

2. Both Q3 and Q4 might be supplied from a common rail/point. So, besides adding Q4 and it's emitter resistor to the CM-piggyback-board, also add the properly decoupled power rail for Q3/4 in terms of another one or two resistor(s), a capacitor and maybe also a zener to the board.

3. Last not least, as another profit, Q4 would provide some uncritical low-impedance connection from this board, to the "regular" VAS on the motherboard.

4. No need to "fiddle" an EF in/onto the main board if there is no original EF yet.

To complete the board, there might be a place provided for Cordell's UHF-stabilizing Cstb and for the two clamping diodes (Cordell's video from 48:20) for better clipping recovery.

There is one slight drawback: This 5-legged FEPCMMPU would make for a far bigger board than your original, nice tiny CM boardies. The 5 connecting points by themselves are not so critical: Of these 5 connections 3 of them would be quite low impadance (Gnd, V- and CM_out). Only the other 2 of them would connect to high impedance points such as the input LTP collectors, which would call for some care when mounting the FEPCMMPU onto the main board.

PS: FEPCMMPU means FunctionallyEnhancedPrecisionCurrentMirrorModPluginUnit.

PPS: And hey - by the way - what does RAP mean? RoastingAtPowerup - or the like?
 

Attachments

  • spend_the_money.png
    spend_the_money.png
    59.5 KB · Views: 152
Hi

Sorry for maybe being some kind of self-indulgent ...

These pictures might illustrate what I meant with the idea of a functionally extended, modder's current mirror board: The circuit, which relates to the aformementionned links, comfortably fits on a 40mmx45mm board using standard tht components.

It includes a helper Q143 (along with C141 for stability at high frequences) for better current and voltage symmetry of the base current mirror Q141/Q142, the Q144 EF for the VAS, two diodes for better clipping recovery, and two cascode Q131/Q132 to provide a GND-referenced, PSRR-friendly connecting point for Cdom. Caveat: The component values are very approximatively chosen and maybe not suitable.

Regards
Simon
 

Attachments

  • ECM_Schem.jpg
    ECM_Schem.jpg
    99.7 KB · Views: 176
  • ECM_Board.jpg
    ECM_Board.jpg
    251.4 KB · Views: 167
  • ECM_3d.jpg
    ECM_3d.jpg
    132.6 KB · Views: 130
Last edited:
Refreshing this excellent thread.
Searching the forum, found this very short thread: ROTEL RA1070

I have this amp, 10+ years from now, and never actually sully satisfied how it does sounds.
On the other hand, the power supply is beaffy, claiming to be good 4Ohms driving ability. MIne is drivig a pair f excellent but power hungry Krell Resolution 3, withut any trouble.
Started with some modification, but not at the extent reported here (not such knoleadgable to go to VAS, etc), summarized as follows:
1. changed op amp (originally OPA2604) to MUSES8920; not bad, a bit more open sound
2. increased capacitor in regulate power supply for op amp (from 470uF to 1000uF, Panasonic FC or FM)
3. Changed all coupling capacitors (10uF/50V Black gate) to Polypropilene 10uF (had on hand Solen, MKP)
4. Increased Greatz from 5A to some bigger, similar type, 25 A

But still not fully satisfied, the bass is unprecise and mid/high still not clear - thus less informative, veiled.

Planned to change and increase the main capacitors, from 1000uF to 15000 or maybe 22000uF/63V, but there is very limited choice since the originals are BHC/ROtel Slit folil and 4-pin shape. The only available caps (mouser) are 22000uF CDE and Kemet (which are the same as in Rotel, BHS is part of Kemet I was told).
So, any further idea=
I think this can be an amp with good basis for making a trully excelent integrated, along side as its predeccessor (RA-985 and RA-980)
 
Member
Joined 2012
Paid Member
Hi,
The RA-1070 has the same power amp circuitry as Paulus' RB-991, so I would suggest you read the posts on what was upgraded in that model (around #446 I believe) - and the audible difference it made.
It is a bit fiddly, but I am working on some new modules, a current mirror CM2 and a VAS4 which should make the process much more straightforward.

First, however, I plan to try it out on a RA-971 to smoke out any problems before offering these new modules to other DIYers.

Ok, yes I know that it has been on my to-do list for a very long time now (in fact over a year!), but other projects always seem to get in the way.:(
So, I guess that y'all just have to be patient a bit longer...

Best,
Per