It's not "so much" gain, it's just right gain. I have no transformers, I have tubes only. And I have about 2,5V of signal level at output.
Yes, I tried as low as 3 ohms per leg. And it played very well, but required HUGE gain at output stage. With 150 ohms per leg it plays outstanding. The sound is crystal clear, no sign of ANY distortion. I think if You have distorted sound and buzz Your PCM1798 chip must be out of order. Frankly I tried also 250 ohms per leg for a while, and it sounded quite OK, but too loud.
Not qiute right. I tried various loads, and 150 ohms is IMHO very good compromise between load to the DAC and required gain of a tube stage.
Yes, I had 1:1, 10k:10k transformers for desimmetrization followed by cathode follower to achieve low output resistance.
Did you just swapped the CS4397 in or there is a need for mods, I removed the output stage and I have lundhal trafos instead.
Best
Pietro
Yes, there is need for mods. For CS4397 I removed the output stage and connected transformers right to the DAC.
And for PCM1798 I removed output stage, and from PCM1798 desoldered everything but DAC. J connected passive I/V and tube gain stage. And PCM1798 sounds far better than any other DIY DAC I ever heard.
The PCM1794 I didn't tried yet.
I also use transformers on the 4398, so I just need to put in the 4397, I thing I'll give it a try
I believe there is some sonci signature in the BB chips, they tend to sound excellent on the bass but I don't like their top midrange.
Best
Pietro
In my opinion you generalize too far. Burr-Brown makes so much DAC chips in various technologies, there are ultra low budget chips and top class chips also. No way to speak about common sonic signature.
Oh bugger.... I just purchased a board as I thought it could natively run higher rates on the USB... I must have been half asleep when I read this thread!
You could use a HiFace M2Tech:
http://www.m2tech.biz/index-3.html
USB from the PC / MAC to S/pdif at 24/192
Hello McGyver
I found the CS4397 on the ebay. One thing I noticed is that while the CS4398 board has switches to select the appropriate sample rate and bit-type, this is not the case with the CS4397. Where did you get your CS4397?
Best Wishes
Pietro
Calling transformer experts
He gave up on me when I asked this over at AA so maybe somebody here will know. What are the inherent design trade offs are when using a higher gain transformer versus a lower one such as when choosing between 1:4 or 1:8 for I/V. It would seem that In the case of the PCM1798/94 where distortion rises rapidly with increased voltage on the outputs pins of the chip that using higher transformer gain with smaller resistances and lower output levels, down to the point where you have just enough to drive your amps to reference listening level, would be the way to go. Is a high gain transformer harder to maintain flat band width? Disadvantaged in current output/ output impedance? I know that they will tend to magnify the channel imbalance caused by the load that the secondary sees but I think I can work around this with careful matching or even with some trimming of the amp termination if need be.
.
The LL1678 and 9206 look interesting since they can accept .1V input which should be enough for my application and the multiple input windings will be ideal to isolate the pairs of outputs in mono mode from offset currents and still have a center tap available.
.
http://www.lundahl.se/pdfs/datash/1678.pdf
.
http://www.lundahl.se/pdfs/datash/9206.pdf
He gave up on me when I asked this over at AA so maybe somebody here will know. What are the inherent design trade offs are when using a higher gain transformer versus a lower one such as when choosing between 1:4 or 1:8 for I/V. It would seem that In the case of the PCM1798/94 where distortion rises rapidly with increased voltage on the outputs pins of the chip that using higher transformer gain with smaller resistances and lower output levels, down to the point where you have just enough to drive your amps to reference listening level, would be the way to go. Is a high gain transformer harder to maintain flat band width? Disadvantaged in current output/ output impedance? I know that they will tend to magnify the channel imbalance caused by the load that the secondary sees but I think I can work around this with careful matching or even with some trimming of the amp termination if need be.
.
The LL1678 and 9206 look interesting since they can accept .1V input which should be enough for my application and the multiple input windings will be ideal to isolate the pairs of outputs in mono mode from offset currents and still have a center tap available.
.
http://www.lundahl.se/pdfs/datash/1678.pdf
.
http://www.lundahl.se/pdfs/datash/9206.pdf
As I understand it the DAC chip sounds better with lower ohm load,I guess you have to select teh transformer to get the right outputsignal,or the otherway around..on the bandwith of the trafo,I don´t know.
With my DAC with Pcm1794,I think it has more body in the middlerange,maybe
thats why I think it sounds darker than the 4398.
I migh try with lower load resitors later on to se what happends.
With my DAC with Pcm1794,I think it has more body in the middlerange,maybe
thats why I think it sounds darker than the 4398.
I migh try with lower load resitors later on to se what happends.
Alcoholics Anonymous?
I'm sure no expert but I agree with your presumption. The load impedance must be kept as low as possible. With higher ratio trafos your output impedance will be higher, and good ones are a hell of a lot harder to make. The high price tag comes with the territory. You will be loading the secondaries pretty hard so resonant issues shouldn't be a problem.
Why don't you try contacting BudP directly for some expert advice. He seems to be very helpful.
AA
.
The LL1678 is $80. So $50 cheaper than the LL1674 that everyone uses in that application. It can't do much more than .8V output though.
Audio Asylum.
.
The LL1678 is $80. So $50 cheaper than the LL1674 that everyone uses in that application. It can't do much more than .8V output though.
trafo theory
Hello
there is any good source to understand how to use signal transformers? I am pretty confused (maybe due to my ignorance on the topic).
I had the problem to connect the CS4398 to the LL1690 and despite all loading circuits discussed here, when I asked Per Lundahl how to use his LL1690 he said "simply put a 7k resistor between pins 10 and 9"... and that's is quite different from all options suggested here. So the question is: what's the *correct* theory about signal transformer?
cheers
Pietro
Hello
there is any good source to understand how to use signal transformers? I am pretty confused (maybe due to my ignorance on the topic).
I had the problem to connect the CS4398 to the LL1690 and despite all loading circuits discussed here, when I asked Per Lundahl how to use his LL1690 he said "simply put a 7k resistor between pins 10 and 9"... and that's is quite different from all options suggested here. So the question is: what's the *correct* theory about signal transformer?
cheers
Pietro
Hi sendler,
with my PCM1798 (in the "small" gigawork dac) I use LL1636 which is the same as LL9206 with additionnal shielding. (but labeled as "mic input" instead of "MC input")
You can have a look here :
http://www.diyaudio.com/forums/digi...new-small-diy-gigawork-dac-5.html#post2204527
Wired as 1+1:10, with 15kOhm across the secondary.
It's the same as putting around 75 Ohm to ground on each DAC leg.
I'm very satisfied by this combo, but I have never listened to some very high end DACs...
Last edited:
Hi Pietro,
I'm sure Per knows his products better than anyone on this forum, I would heed his advice.
There is a very good article about audio transformers written by Bill Whitlock on the Jensen Transformers website. It is loaded with useful info, read it a few times until it sinks in.
I'm wondering where you heard that the 1636 is the same as the 9206. The specs are quite different, the 1636 is a larger trafo and is rated at 0db input. The 9206 is rated at -10db input.
The 1636 looks like it would be fine if there are no saturation problems with low frequencies. You should have a little over 2Vrms for output.
Last edited: