[Pars]

Quote:

Originally Posted by multibit16 /img/forum/go_quote.gif I'm afraid Pedja's comon base stage only works with multibit dacs mainly TDA1541 based. I've built his dac (below) I'm very happy with it also recently built the Ezdac. I had a look for the 965BX service manual but cannot find it, you can give the bypass caps a try with the Burson although I doubt it'll totally cure things. have you changed the output coupling caps etc? if we can find the details for your regulation it should give us some idea's what you can do to improve it

Interesting. I'm currently building this DAC (about 70% done or so). Big transformer.. is this one from the group buy on diyaudio? That will be the most difficult thing for me to source.

 

[multibit16]

Quote:

Originally Posted by Pars /img/forum/go_quote.gif Interesting. I'm currently building this DAC (about 70% done or so). Big transformer.. is this one from the group buy on diyaudio? That will be the most difficult thing for me to source.

Hi Pars,

Yes the transformer is from the group buy, its one we had custom wound for this dac, you could try emailing Antrim , the type number is 7451GPP date 5/05 which is printed on the side.
For the 0.1uf bypass cap on the dac chips -15v I use RTX, I've tried and compared quite a few parts in mine

 

[budx3385]

Quote:

Originally Posted by multibit16 /img/forum/go_quote.gif Before trying the bypassing of the op-amp does this unit have 5534 op-amps? if so see if theres a compensation cap across pin 5 and 8. if so try removing it and seeing if the Burson sounds any better, also if you don't mind the occasional pop and click whilst skipping tracks remove the muting transistors on the output (I presume it uses them) there could be around two on each channel, some cdp's use relays which are better. make sure the coupling cap on pin 6 of the op-amp is good quality, a film would be best but these are bigger, if theres a polar cap then try a good non polar, for a film cap you can try a much lower value Try a few simple things first before splashing out on another output stage

OK. There are two really tiny caps (can't read them) that shunt pins 4 and 7 to ground on both opamps, and then there's a semi-large 220 uF BG near each, but I can't tell if it's in series with pin 6 w/o seeing the backside. THe original opamps were 5534s, but no caps across pins 5 and 8 that I can see. SO, given all that, I tried the 0.1 uF WIMA film caps across pins 4 and 7 of each, and shazam, it sounds way better. Much smoother, really dimensional and textured, gobs of detail, and very musical. Almost smooth enough. I guess if I go a bit further, say 0.33 uF, it might be smoother still, yes?

:Your suggestion for further mod is to exchange the BG 220 uF for a film cap of lower value - such as? I see on the soniccraft website that Dynamicaps are recommended, but for a price.

all the best

 

[Pars]

Have you got the service manual/schematics for this? Rotel gives those out for free (send them an email). 220uf sounds more like power supply decoupling (which is also what the Wimas you put in are for) than output coupling, which is where you would want to use film caps if you were to replace them.

Output coupling will act as a high pass filter... blocking DC but allowing frequencies from ~20Hz and up to pass. The -3dB point (crner frequency) is given by the formula f = 1/2 * Pi * R * C (with C in farads... if the 220uf caps are the output coupling caps, then these would be entered as 0.000220). You should see a resistor to ground after the coupling cap. Also note that the above formula gives the -3dB point... the filter will affect frequencies an octave or more above this point, so the goal typically is to put the -3dB point as low as practical... sub-10Hz certainly, and lower.

 

[multibit16]

Quote:

Originally Posted by budx3385 /img/forum/go_quote.gif OK. There are two really tiny caps (can't read them) that shunt pins 4 and 7 to ground on both opamps, and then there's a semi-large 220 uF BG near each, but I can't tell if it's in series with pin 6 w/o seeing the backside. THe original opamps were 5534s, but no caps across pins 5 and 8 that I can see. SO, given all that, I tried the 0.1 uF WIMA film caps across pins 4 and 7 of each, and shazam, it sounds way better. Much smoother, really dimensional and textured, gobs of detail, and very musical. Almost smooth enough. I guess if I go a bit further, say 0.33 uF, it might be smoother still, yes? :Your suggestion for further mod is to exchange the BG 220 uF for a film cap of lower value - such as? I see on the soniccraft website that Dynamicaps are recommended, but for a price. all the best

Pleased to hear the film cap across the +/-v supply pins helped

sure you can try a little higher value, it may not make a big difference but is worth a try if you have the parts already, it may make the bass a little heavier and add some more smoothness, best thing to do is experiment.

Again with the output coupling caps you'll have to experiment, obviously you may be limited with large value film caps because of the size.
What I would do is try a few different value cheap polyester first in say 1uf, then 3.3uf or 4.7uf and maybe 10uf if they will fit, it gives you an idea what value you can get away with, once your happy with the value you can buy a better quality type in the same value.
If your pre/amp already uses coupling caps on the input its worth measuring the DC on pin 6 of the op-amp/burson pin, if the DC offset is very low in the mV you could simply bypass the coupling cap altogether with a wire link but be sure the pre/amp has input coupling

 

[multibit16]

Quote:

Originally Posted by Pars /img/forum/go_quote.gif Have you got the service manual/schematics for this? Rotel gives those out for free (send them an email). 220uf sounds more like power supply decoupling (which is also what the Wimas you put in are for) than output coupling, which is where you would want to use film caps if you were to replace them. Output coupling will act as a high pass filter... blocking DC but allowing frequencies from ~20Hz and up to pass. The -3dB point (crner frequency) is given by the formula f = 1/2 * Pi * R * C (with C in farads... if the 220uf caps are the output coupling caps, then these would be entered as 0.000220). You should see a resistor to ground after the coupling cap. Also note that the above formula gives the -3dB point... the filter will affect frequencies an octave or more above this point, so the goal typically is to put the -3dB point as low as practical... sub-10Hz certainly, and lower.

It does seem high, I think some of the Marantz units had highish value coupling caps although I can't remember if it had two in series making them back to back so halving the value like a bi-polar

 

[Pars]

Quote:

Originally Posted by multibit16 /img/forum/go_quote.gif <snipped> If your pre/amp already uses coupling caps on the input its worth measuring the DC on pin 6 of the op-amp/burson pin, if the DC offset is very low in the mV you could simply bypass the coupling cap altogether with a wire link but be sure the pre/amp has input coupling

Good advice on the film caps... play around with cheap for frequency response, then get Dynamicaps or whatever when you have an acceptable range/physical size nailed down.

My RCD-855 has around 3.3V (not mV) of offset because of the way the opamps are configured (I haven't sat down and analyzed the circuit to see why... going to go discrete anyhow so why bother

). If your Rotel is setup anything like my older one, you are going to have significant DC on the outputs, so don't be surprised. I even tried rolling some FET input opamps in just to see if it was a bipolar vs. FET thing... offset stayed the same.

 

[multibit16]

Quote:

Originally Posted by Pars /img/forum/go_quote.gif Good advice on the film caps... play around with cheap for frequency response, then get Dynamicaps or whatever when you have an acceptable range/physical size nailed down. My RCD-855 has around 3.3V (not mV) of offset because of the way the opamps are configured (I haven't sat down and analyzed the circuit to see why... going to go discrete anyhow so why bother ). If your Rotel is setup anything like my older one, you are going to have significant DC on the outputs, so don't be surprised. I even tried rolling some FET input opamps in just to see if it was a bipolar vs. FET thing... offset stayed the same.

Does the Rotel RCD-855 use TDA1541? most of the cdp's using the multibit dacs have high DC offset after the I/V op-amps, some don't even bother using a nulling circuit after the dac chip, I have a modified Arcam Alpha5 that uses a nulling circuit after the dac chip before it goes to the I/V op-amps input but theres still around 1v DC after the filter stage op-amps, I use 1uf PPS film coupling caps in that with good results

 

[Pars]

Quote:

Originally Posted by multibit16 /img/forum/go_quote.gif Does the Rotel RCD-855 use TDA1541? most of the cdp's using the multibit dacs have high DC offset after the I/V op-amps, some don't even bother using a nulling circuit after the dac chip, I have a modified Arcam Alpha5 that uses a nulling circuit after the dac chip before it goes to the I/V op-amps input but theres still around 1v DC after the filter stage op-amps, I use 1uf PPS film coupling caps in that with good results

Yeah, its the std. Philips SAA7210/SAA7220/TDA1541A. Stock opamps were NE5532s. I've rolled AD826s, LM6172s and currently LM4562s in it, as well as a couple of others. Opamp rolling provided some amusement, but nothing that I would calll stellar in terms of performance... most of the competent ones sound pretty similar... better than stock however. No nulling circuit, stock it used 100uf BG stds. as both coupling caps and PSU decoupling around the opamps... no ceramics or films around the opamps. I guess for the time (1990) and price, pretty good components and design. PSU has +/-5 and +/-15V available. Replaced the two bridges (stock 1N4007) first with UF4007s and now with 11DQ10 Schottkys. I replaced the output coupling caps with BG NX 47uf, helped some. Removed the muting transistors... not sure if I hear alot of improvement with that. I do notice it alot if I forgetfully power the CD player up with the preamp unmuted, however

I did not try any film caps. Tentlabs XO module on a Flea-based clock (home etched modified board with /2 circuit to directly clock DAC and get the 7220 out of the colck distribution role as it is supposedly extremely noisy and jitter prone).

 

[budx3385]

Great ideas and comments - thank you very much, gentlemen. I just sent a request to Rotel for a schematic. I have ordered 0.33 and 0.47 uF bypass caps. I will order next some larger film caps to try in the coupling path. I have not yet figured out how to disengage the muting transistors - I see a discrete pair on each channel just before the RCA output,, and I don't know whether those are a buffer or "muting transistors".

 

[Pars]

Muting transistors: Normally they just short the outputs to ground... to disable them you just remove them. Better yet would be to replace these with a relay circuit. Most people can live with them disabled... I don't hear any noise when doing a search, etc. on a CD. The only problem is if you power the player up or down with amp volume up and unmuted.

 

[multibit16]

Quote:

Originally Posted by Pars /img/forum/go_quote.gif Yeah, its the std. Philips SAA7210/SAA7220/TDA1541A. Stock opamps were NE5532s. I've rolled AD826s, LM6172s and currently LM4562s in it, as well as a couple of others. Opamp rolling provided some amusement, but nothing that I would calll stellar in terms of performance... most of the competent ones sound pretty similar... better than stock however. No nulling circuit, stock it used 100uf BG stds. as both coupling caps and PSU decoupling around the opamps... no ceramics or films around the opamps. I guess for the time (1990) and price, pretty good components and design. PSU has +/-5 and +/-15V available. Replaced the two bridges (stock 1N4007) first with UF4007s and now with 11DQ10 Schottkys. I replaced the output coupling caps with BG NX 47uf, helped some. Removed the muting transistors... not sure if I hear alot of improvement with that. I do notice it alot if I forgetfully power the CD player up with the preamp unmuted, however I did not try any film caps. Tentlabs XO module on a Flea-based clock (home etched modified board with /2 circuit to directly clock DAC and get the 7220 out of the colck distribution role as it is supposedly extremely noisy and jitter prone).

For common op-amp based I/V the THS4032 work quite well if you can get them to behave, they need a really good supply though to stay stable, I still like pedja's common base I/V though

The Flea based clocks are really good, sounds like we both did the similar thing

For my Arcam A5 the XO feeds the SAA7310 direct with a 1mohm resistor going across the SAA7310 XO/XI, I then built a separate divider circuits to directly feed the TDA1541A's from the Flea, I used another Flea to power the dividers.
It made quite a big difference tbh, I also used another modified flea to run the dacs +5v and separate regs for the -5/-15v section of the dac, Burson regs was used for the I/V section.
The TDA1541's really come to life with clean supplys, if you still use oversampling its best to have the SAA7220 on its own supply, that chip certainly is noisey! it injects lots of crap into the supply which isn't good if the dacs +5v is on the same circuit

 

[multibit16]

Quote:

Originally Posted by budx3385 /img/forum/go_quote.gif Great ideas and comments - thank you very much, gentlemen. I just sent a request to Rotel for a schematic. I have ordered 0.33 and 0.47 uF bypass caps. I will order next some larger film caps to try in the coupling path. I have not yet figured out how to disengage the muting transistors - I see a discrete pair on each channel just before the RCA output,, and I don't know whether those are a buffer or "muting transistors".

No problem, glad I could help

Let us know how you get on.
I agree with Pars, I'd say those transistors are for the muting, simply remove them!

 

[Pars]

Quote:

Originally Posted by multibit16 /img/forum/go_quote.gif For common op-amp based I/V the THS4032 work quite well if you can get them to behave, they need a really good supply though to stay stable, I still like pedja's common base I/V though The Flea based clocks are really good, sounds like we both did the similar thing For my Arcam A5 the XO feeds the SAA7310 direct with a 1mohm resistor going across the SAA7310 XO/XI, I then built a separate divider circuits to directly feed the TDA1541A's from the Flea, I used another Flea to power the dividers. It made quite a big difference tbh, I also used another modified flea to run the dacs +5v and separate regs for the -5/-15v section of the dac, Burson regs was used for the I/V section. The TDA1541's really come to life with clean supplys, if you still use oversampling its best to have the SAA7220 on its own supply, that chip certainly is noisey! it injects lots of crap into the supply which isn't good if the dacs +5v is on the same circuit

I just got a few boards made (see this thread), 2 of which were the modified Flea boards I laid out (see pinkfishmedia Got Fleas thread... these were my last schematic posted with another opamp PSU onboard to power the 74HC logic seperate from the clock), as well as 3 standard Flea boards which I laid out in Eagle to get better trace widths/clearances for home etching.

Sounds like you have really gone extensive on your PSU mods... I may have to try powering the TDA1541A with its own flea. Are you using Pedja's common-base I/V? I have a Pedja v1.1 DAC board that I am building that has these on it, but its going to be awhile before I finish it. I may want to build an I/V board to use internally in the Rotel... any info you wish to share would be appreciated.

Hope you don't mind the thread-crapping Bud

Chris

 

[multibit16]

Quote:

Originally Posted by Pars /img/forum/go_quote.gif I just got a few boards made (see this thread), 2 of which were the modified Flea boards I laid out (see pinkfishmedia Got Fleas thread... these were my last schematic posted with another opamp PSU onboard to power the 74HC logic seperate from the clock), as well as 3 standard Flea boards which I laid out in Eagle to get better trace widths/clearances for home etching. Sounds like you have really gone extensive on your PSU mods... I may have to try powering the TDA1541A with its own flea. Are you using Pedja's common-base I/V? I have a Pedja v1.1 DAC board that I am building that has these on it, but its going to be awhile before I finish it. I may want to build an I/V board to use internally in the Rotel... any info you wish to share would be appreciated. Hope you don't mind the thread-crapping Bud Chris

First well done on the PCB's Chris, you've made a lovely job of those!
I see you have a AD815 using MickF's layout, I built one of those too!
You know I come on headfi and I haven't even got a decent headphone amp

thats the next thing I need to build, its the main reason I came on here to get some idea's, I mainly use my main hi-fi but now I'm desparate for something nice to allow late night listening, my Chiarra and Millet amp isn't too bad but not a patch on the quality I get with main hi-fi

So any advice there would be appreciated, its hard getting pcb's in the UK.

Anyway (sorry Bud from me too) I have managed to implement Pedja's common base and nulling circuit on the Arcams pcb, it wasn't easy and required some linking out etc in various places but I did manage to do it, I'm very happy with the sound, the Arcam sounds 100% better than it was in stock form.
You should be able to implement it in your Rotel, it may be a bit harder if the Rotel was designed for dual op-amps though.
With the Clocking,better supplys and I/V stage you should get a big improvement