A little story on my progress in TDA1541A DAC building.
I modded two Red Baron PCB to accept 50 Hz DEM mod and used them with Ian Canadas I2S to PCM converter in balanced mode and with passive I/V converter. This immediately bettered my, uptil now , preferred DAC with Ian Canadas FIFO and dual clock with isolator and custom designed I/V converter.
I had to dismantle it again as it was kind of a rats-nest construction and I want to put it in a nice box with remote vol. control etc.
So I had a third red baron pcb and I just got two pcb from ryanj with ecdesigns version of a I2S to Simultaneous converter (worked perfectly first time , thanks ryanj!) and I wanted to make a dac I could use , while building the balanced dac (it also needs to have 4 channels, as I want to make a digital corssover for future subs).
The dac was unmodded (besides 1uF acrylic SMD capacitors for DEM decoupling) and had the original single MOSFET I/V converter. Hmmm good but not great. Then I switched to passive I/V converter and I got a step closer, but not enough. I then modded it with 50Hz DEM and , wow, what a difference! This is a very good mod especially when you run the TDA1541A in simultaneous mode... Spaciousness and perspective is much better and the there is black darkness in the space between and behind instruments, so you can easily hear the recorded rooms response to the instruments..
The output from the passive I/V is too low for my amps , and I did not have a good enough preamp to deal with that, so I desided to try to mod. my TDA1543 I/V converter to work with TDA1541A. It was rather an easy mod and now I had plenty of output to my amps. Comparing the passive I/V to the modded active I/V, @ the levels that the passive could obtain, showed that the active was at least on pair with the passive solution. I had to turn down the digital vol quite a lot for the active I/V and I get lesser resolution that way. When I turn up the vol. the active performs even better.
The active showed me, when the digital vol was turn way down, that there was some noise overlaying the weak signals. Most of it was probably quantization noise, but when I added 100nf smd parallel to the 100 uF DEM capacitors, it was reduced, so I would recommend that.
Both the MOSFET I/V and the modded I/V (also a non-feedback circuit) have way lower distortion figures than the passive solution (FFT analyzer result)....Makes one wonder ...........
Concerning the I/V stage: I like MOSFETS and especially JFets as input devices and prefer them to bipolar, but in the event of an I/V converter bipolar are IMHO much better. They shine when they are used as current operated devices (that is their nature).
I hope this can be used as inspiration for others, but it is only my personal observations and of course subjective..
Hi Koldby,
Could you please upload some pictures, how you modified the Red Baron pcb? I would like to avoid modding my one and only RB5 pcb, thats the reason why I go for Ryanjs pcb. But if the modification looks easy maybe I try it.
Which type of 100uF elko are you using?
Modding the Red Baron pcb is very easy. Just put 4 of the 100/25 v caps on the component side and the rest , 3 pcs, on the copper side. There are exactly room for soldering the caps. The caps are Nichicon KA series .
If you have a newer Red Baron, there are jumpers for changing it to simultaneous mode, else you have to cut some traces here and there
Attachments
Thank you, Koldby!
I have the latest version of RB, the 5th edition, I can switch to simultaneous mode with jumpers. But the most important difference is that the dem capacitors intended to be only smd parts, therefore there are no wholes for the pins and the "copper" side can not be used to accommodate capacitors. So there is no place for the 100u-s. Have to seek another option.
I have the latest version of RB, the 5th edition, I can switch to simultaneous mode with jumpers. But the most important difference is that the dem capacitors intended to be only smd parts, therefore there are no wholes for the pins and the "copper" side can not be used to accommodate capacitors. So there is no place for the 100u-s. Have to seek another option.
Sounds like you're getting current leakage through the oscons. You'll need to use a low leakage cap like the Nichon KA series with a voltage rating of 25V.
I met the similar situation as I modified the CAP at analog region(decouple or couple)
I used to use the NOS caps like roe sic-safco......thus I blame for it's age.
But the sound will getting better and better with burn in.
There is once that is so worse that I thought it might be broken somewhere,but after 2 more
days the phenomenon went away and never happened.
Maybe it's totally different situation here
But I do agree the leakage is the root cause of it
I used to use the NOS caps like roe sic-safco......thus I blame for it's age.
But the sound will getting better and better with burn in.
There is once that is so worse that I thought it might be broken somewhere,but after 2 more
days the phenomenon went away and never happened.
Maybe it's totally different situation here
But I do agree the leakage is the root cause of it
You're right, they definitely have lower leakage (3uA vs 0.2uA at rated voltage).
I used UKA because this is what John recomended:
An externally hosted image should be here but it was not working when we last tested it.
Can’t wait for the 100uF
So try to modify to0.1uF and keep the0.1uF first.
The change is obviously.but the treble is a little bit too much
Maybe cause by the decouple caps can’t follow
Echo and harmonics are completely and neat!
Last edited:
I had the same issue initially. It has gone after a couple of hours. I used Nichicon Muse 100uF/25V. Pictures are several pages back.
I've just been comparing (on paper) a number of caps that are immediately available to me as I would have to wait on the Nichicon KA. So can confirm there are many with about the same leakage as the KA eg. Panasonic FM & NHG, Nippon Chemicon KY, Rubycon YXJ, Elna RE2 & RJB
The KL are larger diameter (8mm cf 6.3mm for KA) so you can not group them as close to the pins; also ripple current is higher for KL.
•Still something confuse me. I found that EC change to use 50Hz about #6069.
Before that he try the higher freq DEM which I can get the point then.
The DEM accuracy seems to be better with high freq oscillator. But the low freq. modify do get better, why??
That he mention about facing some “DEM problem". But I wonder to know why it won’t cause some effect as it drop into the audible region.
I know this might discussed before but I’m sorry that I just can’t find it.
the other question is it seems to use two resistor at pin16 and pin17 connect to pin15
why change two a single registor?
Before that he try the higher freq DEM which I can get the point then.
The DEM accuracy seems to be better with high freq oscillator. But the low freq. modify do get better, why??
That he mention about facing some “DEM problem". But I wonder to know why it won’t cause some effect as it drop into the audible region.
I know this might discussed before but I’m sorry that I just can’t find it.
the other question is it seems to use two resistor at pin16 and pin17 connect to pin15
why change two a single registor?
Last edited:
Probably have too much leakage.
I think this is very important since the currents involved are very small and this is proved by the need to wait several minutes to charge up 100uf.
UKL are the lowest leakage ones I could find and they are larger diameter but since the dem frequency is 50Hz instead of 175-200Khz, I do not think the need for longer tracks to them is a problem.
My dual 1541 pcbs will arrive at the weekend so I hope to report back next week on the results with the UKL caps.