PCM56 DAC - Power Supply Question

Member
Joined 2004
Paid Member
I'm a fan of transformer I/V for current output DAC chips.

I tested pure resistors, some transistor or FET based I/V, but more than a decade I use Sowter 8347 for my PCM63 DAC, with only secondary load (2k). It results the most engaging sound for me.

But no free lunch, must to use a gain stage to produce 2V RMS "standard" output at "0dB" digital input.

In those days I used tube stage, Siemens CCa SRPP ala Audio Note.

Another alternative is using (DHT) tube with active load (for example "gyrator") to achieve lower output impedance.
 

Attachments

  • PCM63 transformer I_V.jpg
    PCM63 transformer I_V.jpg
    90.3 KB · Views: 250
Member
Joined 2019
Paid Member
Hey euro21 - I considered much of the above for this design! I decided however to use all passive I/V with the Sowter SUT. Glad you find the I/V transformer best, hopefully that means I am on a good path.

I've had a change of heart on one post above - I've decided to buffer the outputs of the digital front end module I am using (I2SoverUSB from JLSounds) and run four PCM56 in parallel per channel.

This will give +/- 4mA, then will use 2K-3.3K on the secondary of the 1:10 SUT. With 3.3K, output voltage will be roughly 0.85Vrms and I think I will live with the lower output voltage without an additional gain stage. Will try this first.

Looking at HCMOS buffer chips now, perhaps SN74HC126N? We will see.
 
Last edited:
Member
Joined 2019
Paid Member
Another alternative is using (DHT) tube with active load (for example "gyrator") to achieve lower output impedance.

Are you still using this type of I/V stage, euro21? Transformer + tube gain? I am revisiting the idea again, this was my original idea before planning for fully passive I/V with only transformer for gain. The idea seems simple, but to get reasonable output voltage and output impedance balanced, it is getting more complex to parallel many chips.

I like the idea of using DHT of course, but the power supply is larger, concerned about noise pickup from filament transformers.

One IDHT option I was just looking over is 12B4A with battery bias (2x 9VDC in series), and choke or active load. Has mu of 6.5 and 1K plate resistance, could achieve 2Vrms with 1:5 SUT with 2.5K secondary resistor using a single PCM56 per channel.
 
Last edited:
Member
Joined 2004
Paid Member
Yes, this is my PCM63 DAC output buffer.

The output impedance a little bit high, but I use 801 preamp, so it's acceptable.

p.s. the R100 value is typo, the real value is 2k.
 

Attachments

  • PCM63, transformer I_V, tube buffer.jpg
    PCM63, transformer I_V, tube buffer.jpg
    49.3 KB · Views: 205
Last edited:
Member
Joined 2004
Paid Member
I tried 5842, 12B4a, 6N30P(and DR version) for my friend, with Noguchi 10k:600 OPT.

The first is costly game, the second (from my trash) is problematic (most of them microphonic), the last is good.

The DR version was the best, but it's more expensive now than good 5842. :-(
 
Member
Joined 2019
Paid Member
Hmmm, thank you, I have to look at some options, this is becoming appealing again vs. passive with SUT, although I am very curious to know what that would sound like.

I thought I had found a good choice in 12B4A, but microphony is no good. Have you found there is an advantage in higher vs. lower gain tube and adjusting the transformer secondary load? My thought was to use low gain triode with low plate resistance and increase the SUT secondary load to achieve 2Vrms output.

The reason I ask, I still have many E55L that I am not using. Mu is a little high at 30, but a single PCM56 with perhaps 560ohm secondary would give 2Vrms output, primary seen by PCM56 would only be 20ohms or so. Output impedance would be only 600ohm this way. Would probably use a cooler bias for E55L for longevity...swinging only 2Vrms, could bias something like 80V / 55mA.

NmQJWTZ.png


Just one idea, but I do like it!
 
Last edited:
Member
Joined 2004
Paid Member
Some years ago I tested E55L (80V, 22ma) with 1:1 interstage, but rather it was a driver than preamp (more than 27dB gain).

We have a local proverb to exaggeration : "shooting to sparow with cannon". :p

I preserve this tubes for building -for example- good phono stage.

Try to search tubes with 8-10x gain, and low distortion.
 
Member
Joined 2019
Paid Member
Yes but if the sparrow is completely obliterated my ears will rejoice :)

You are probably right, I should save for LCR phono or the like.

Tubes with 8-10x gain and low plate resistance and good linearity was exactly what I was looking for when I came across 12B4A. I will continue my search for the ideal...
 
Member
Joined 2019
Paid Member
That could work! Perhaps run one section as VAS with gyrator load direct-coupled to section 2 as cathode follower.

Using the same idea, could use a dissimilar dual triode like 6EW7 - section 1 mu 17.5 with section 2 as CF - output impedance below 200ohm.
 
When it comes to tubes in the output stage of the DAC, I recently inserted 6HM5 into my DAC, the gain of the first stage is around x70, the second stage is WCF with an output resistance below 30ohm.
I used to use the 6S4P for the first stage before, however I only got the 6HM5 recently and am very happy with the sound.
 

Attachments

  • grunf DAC 1.png
    grunf DAC 1.png
    51.1 KB · Views: 217
  • 6HM5 tube I-V converter 3.jpg
    6HM5 tube I-V converter 3.jpg
    419.2 KB · Views: 190
  • Like
Reactions: 1 user
Member
Joined 2019
Paid Member
Very nice grunf, interesting circuit. Tube active I/V is something I would like to try in the future. Have you compared to other types (passive? Si active stages?).

Right now I am planning to use Sowter SUT for I/V conversion. Looking at a tube gain stage afterward.

I am thinking either dual triode with gain stage direct-coupled to CF, or I might keep it as simple as possible and use a single triode common cathode gain stage.

This is what I am interested in at the moment, simple 12B4A gain stage, battery biased with CCS load after the 1:5 SUT. Could also try in 1:10 with higher secondary resistor (1/2 load at the PCM56 output).

rLNAuCH.png


Very simple, but that's why I like it! Since it is only swinging 2Vrms, can use a low voltage, high current bias point, which makes the output impedance ~700-800ohm. My amplifiers have typical input impedance of 50K, some 25K, so I think it will work well.

Very curious if you have compared tube I/V to other types, please let me know :D
 
I tried everything except the transformer, from passive to op.amp I / V stage and finally I discovered a grounded grid for I / V. The op.amp is good, but with a quality power supply and with a 'slightly' more expensive op.amp. I used AD811 with Walt Jung superreg. and achieved good results with TDA1541 and PCM1702. I then switched to passive with tubes, initially with ECC88 and later 6S4P.
The discovery of the grounded-grid with the 6S4P forever left the passive stages behind me. I got a sound full of dynamics, details, so to speak, a real discovery. The sound of the passive stage is sluggish, it lacks those dynamics and speed.
The 6HM5 is great as an I / V stage, even better than the 6S4P. Great tube, I recently bought a dozen pieces and could hardly wait to put them in my DAC.
With this stage I directly drive the amplifier over a 10K Penny & Giles sliding pot. The output voltage is about 5Vpp.

One recommendation, for M2 DN2540 is not good for this position, try more fet in parallel or BSS159N for higher currents. I mostly use 2SK3557-7 in parallel, I recently tried BSS159N as CCS and it is good.
 
Member
Joined 2019
Paid Member
I tried everything except the transformer, from passive to op.amp I / V stage and finally I discovered a grounded grid for I / V. The op.amp is good, but with a quality power supply and with a 'slightly' more expensive op.amp. I used AD811 with Walt Jung superreg. and achieved good results with TDA1541 and PCM1702. I then switched to passive with tubes, initially with ECC88 and later 6S4P.
The discovery of the grounded-grid with the 6S4P forever left the passive stages behind me. I got a sound full of dynamics, details, so to speak, a real discovery. The sound of the passive stage is sluggish, it lacks those dynamics and speed.
The 6HM5 is great as an I / V stage, even better than the 6S4P. Great tube, I recently bought a dozen pieces and could hardly wait to put them in my DAC.
With this stage I directly drive the amplifier over a 10K Penny & Giles sliding pot. The output voltage is about 5Vpp.

Very interesting, thank you grunf. I will take this with me, going to go through with my SUT + tube gain stage approach for the first time. With some experience under my belt, I will try grounded grid active I/V next. I think I've come across some papers on the design, there is an article on TubeCAD I believe, will do some reading on the topic.
 
Hi,


which ECC88 tubes have tried with the tda1541, please ?

I tried a Ei ,Tesla and rusian 6N23P and 6N1P
Ei has the best sound, followed by the Russian 6N1P.
Ei tubes are made under the license of Philips, even the box is the same.
In the box in the picture are the Ei PCC88 and 6HM5 I recently bought, so I now have a good stock.
Today, it is almost impossible to get more Ei ECC88 or PCC88, and the price goes up.

Which sort of tube circuitry do you use for the TDA 1541 ? Mu follower ?


Interesting testimonial you made...

With ecc88 initially classic SRPP, and later anode amp. with CCS, due to small amplification afterwards I switched to 6S4P with CCS. CCS was made with DN2540 as in the scheme a couple of posts before and plays an essential role in the sound itself. DN2540 I later replaced with IXTP08N100.
 

Attachments

  • 01.jpg
    01.jpg
    229.6 KB · Views: 112
  • WCF with Ei ECC88.jpg
    WCF with Ei ECC88.jpg
    227.3 KB · Views: 121