i'm currently looking for a cheap 1:20 step up trafo to use with a differential 1541 dac that i'm planning. it will have passive i/v then connected directly to the trafo for bal->unbalanced output.
my budget is terribly low at the moment.
Cinemag CMQEE-3440A seems like a winner but $88 a piece is still too expensive.
i'd like to try the circuit out with much cheaper trafos first.
any suggestions?
i can also custom wind them to a desired spec. they would be cheaper than 3440a, too.
my budget is terribly low at the moment.
Cinemag CMQEE-3440A seems like a winner but $88 a piece is still too expensive.
i'd like to try the circuit out with much cheaper trafos first.
any suggestions?
i can also custom wind them to a desired spec. they would be cheaper than 3440a, too.
I suggest to study step up transformer operation thoroughly before applying one in your DAC. Following link might be useful:
mc step-up transformers explained
There are many drawbacks to what you hope to do. Perhaps, the first technical issue to realize is that a 1:20 step-up transformer will multiply the effective I/V source resistance by a factor of 400. For example, if you implement a pair of 5 ohm balanced I/V resistors, the output impedance from the transformer's secondary will be at least 4,000 ohms (5 ohms * 2 * 400), which is quite high. You could reduce the value of those I/V resistors, but then you are reducing the resulting signal amplitude as well. There are many other issues presented by this approach as well, not the least of which is the goal of low cost. I strongly suggest that you drop your plan to utilize a passive output based only on a high step-up ratio transformer.
Last edited:
As you noticed, step-up transformers are expensive... especially good ones.
Ken Newton is imho right in his advice to you.
I also think you'd better use good opamps or an other type of discrete I/V (transistor or tube based). from my perspective, transformers are only good in power-supplies, not in the signal-path.
Ken Newton is imho right in his advice to you.
I also think you'd better use good opamps or an other type of discrete I/V (transistor or tube based). from my perspective, transformers are only good in power-supplies, not in the signal-path.
Last edited:
Mouser now stocks PQ type ferrite cores, which at least are fairly cheap. I've never tried a step-up trafo for a DAC, preferring to use active components. But I've wound various step-down transformers on PQ cores and got pretty good results sonically. The best way to learn is to have a play - notwithstanding the caveats mentioned by Ken. If you do want to go for step-up, I'd suggest a more modest turns ratio at first, say 1:5. You'll need to invest in a reel of fine wire (say from 0.1mm to 0.2mm diameter) for the winding with the most turns.
6698272021 Fair-Rite | Mouser
6698272021 Fair-Rite | Mouser
thanks for the suggestions. i decided to give cen i/v a shot, followed by 600:600 trafo for unbalancing.
also might try passive i/b >op amp gain just for comparison.
my past experience with ad844+ 1543 was that though 844 has far superior parameters to passive i/v (on 1543), it doesn't necessarily translate to musical enjoyment.
i guess i'm one of those 'if it sounds good...' types.
also might try passive i/b >op amp gain just for comparison.
my past experience with ad844+ 1543 was that though 844 has far superior parameters to passive i/v (on 1543), it doesn't necessarily translate to musical enjoyment.
i guess i'm one of those 'if it sounds good...' types.
gltch, noise and tonal balance...
hi PRESAPIAN,
If trafo is no of a good quality like a Lundhal, the sound would be better without... better a good cap than a bad trafo. A good trafo can be less expensive than a very good cap (Teflon, oil). But if you use tube a very very good trafo will be more expensive than the best caps.
With the design maid with no dc blokers, adding caps can resolve sometimes problems by transforming tonal balance for a best match with the rest of your system. For example some old green Nichicon Muse bipolarized are fantastic (far better than the new one, don't know why).
You can test two serie SIlmic or Silmic 2 in + - - + arrangement. MKP can be worse, Teflon are too expensive but the russian.
Saying than no caps is the best cap is too simple and not always true...depend on the setup and hifi sytem and how you manage the output cap to use as "tonal rectifier" and high pass filter. Sometimes you can have better bass with caps because of the final band pass they "add" at the output of the source in relation to the rest of the system.
There is here a german fellow which use speakers filter after PCM56 dac chip with good results but never try myself.
Of course IMHO, I'm not a designer nore a specialist.
hi PRESAPIAN,
If trafo is no of a good quality like a Lundhal, the sound would be better without... better a good cap than a bad trafo. A good trafo can be less expensive than a very good cap (Teflon, oil). But if you use tube a very very good trafo will be more expensive than the best caps.
With the design maid with no dc blokers, adding caps can resolve sometimes problems by transforming tonal balance for a best match with the rest of your system. For example some old green Nichicon Muse bipolarized are fantastic (far better than the new one, don't know why).
You can test two serie SIlmic or Silmic 2 in + - - + arrangement. MKP can be worse, Teflon are too expensive but the russian.
Saying than no caps is the best cap is too simple and not always true...depend on the setup and hifi sytem and how you manage the output cap to use as "tonal rectifier" and high pass filter. Sometimes you can have better bass with caps because of the final band pass they "add" at the output of the source in relation to the rest of the system.
There is here a german fellow which use speakers filter after PCM56 dac chip with good results but never try myself.
Of course IMHO, I'm not a designer nore a specialist.
Last edited:
I have built a few output stages for TDA1541A, and the best is this one:
http://community.fortunecity.ws/rivendell/xentar/1179/projects/adagio/Adagio.html
This circuit uses a cheap Amplimo TM-3 transformer:
TM3 step-up transformer 1:15
We could discuss about the electrical properties of transformers, about impedances, transfer functions, core saturation, distortion, ... but facts are facts. And this is one of the best DACs I've ever heard, and way the best of those based on TDA1541A.
http://community.fortunecity.ws/rivendell/xentar/1179/projects/adagio/Adagio.html
This circuit uses a cheap Amplimo TM-3 transformer:
TM3 step-up transformer 1:15
We could discuss about the electrical properties of transformers, about impedances, transfer functions, core saturation, distortion, ... but facts are facts. And this is one of the best DACs I've ever heard, and way the best of those based on TDA1541A.
If I -had- to make a -fully- passive I/V stage for the TDA1541A I would do following:
1) Resistive divider that provides +240mV. For example 24 Ohms in series with 476 Ohms connected to the +5V supply and a suitable electrolytic cap (1000uF/6V3 for example) in parallel with 24 Ohms.
0.002mA bias current results in 240mV drop across a 120R resistor. Without further action we would end up with -0.24V DC offset on the I/V resistor and on the connected transformer. By lifting the I/V resistor GND reference to +0.24V this DC offset voltage is nulled (-0.24V) + (+0.24V) = 0. And this can still be achieved with passive components (2 resistors and a capacitor).
2) Connect a 120R Welwyn W21 wire wound resistor between +240mV and TDA1541A output. With TDA1541A full scale current of 4mA this results in 480mVpp. This is still below the clipping limit of around 600mVpp and distortion stays below the audibility threshold of roughly 1%. It is -very- important to use a low noise wire wound (not bulk metal foil !!) resistor here. Welwyn W21 and Rhopoint econister are recommended.
3) Connect a good quality 1:4 step up transformer to the 120R resistor (LL1674 for example). Use a transformer that works optimal with a source load of around 100 … 200 Ohms.
4) Connect a RC filter to the transformer secondary to suppress ringing (use test signal and scope to tune).
You will then have a fully passive I/V stage that provides 1.92Vpp and an output impedance of approx. 1900 Ohms (120R * 4 * 4).
Low passive I/V resistor values give lowest distortion but poor resolution as most of the LSBs are burried deep below the noise floor.
With 12.5 Ohm passive I/V resistor (50mVpp) LSB is only 0.05 / 65536 = 762 nanovolts. Noise levels of up to 1 millivolts in an I/V stage are quite common. So 6 MSBs will stay above the noise floor while the other 10 hide below it. Suppose one could still hear signals that are below the noise floor one might just be able to squeeze out 8 bits resolution ……
1) Resistive divider that provides +240mV. For example 24 Ohms in series with 476 Ohms connected to the +5V supply and a suitable electrolytic cap (1000uF/6V3 for example) in parallel with 24 Ohms.
0.002mA bias current results in 240mV drop across a 120R resistor. Without further action we would end up with -0.24V DC offset on the I/V resistor and on the connected transformer. By lifting the I/V resistor GND reference to +0.24V this DC offset voltage is nulled (-0.24V) + (+0.24V) = 0. And this can still be achieved with passive components (2 resistors and a capacitor).
2) Connect a 120R Welwyn W21 wire wound resistor between +240mV and TDA1541A output. With TDA1541A full scale current of 4mA this results in 480mVpp. This is still below the clipping limit of around 600mVpp and distortion stays below the audibility threshold of roughly 1%. It is -very- important to use a low noise wire wound (not bulk metal foil !!) resistor here. Welwyn W21 and Rhopoint econister are recommended.
3) Connect a good quality 1:4 step up transformer to the 120R resistor (LL1674 for example). Use a transformer that works optimal with a source load of around 100 … 200 Ohms.
4) Connect a RC filter to the transformer secondary to suppress ringing (use test signal and scope to tune).
You will then have a fully passive I/V stage that provides 1.92Vpp and an output impedance of approx. 1900 Ohms (120R * 4 * 4).
Low passive I/V resistor values give lowest distortion but poor resolution as most of the LSBs are burried deep below the noise floor.
With 12.5 Ohm passive I/V resistor (50mVpp) LSB is only 0.05 / 65536 = 762 nanovolts. Noise levels of up to 1 millivolts in an I/V stage are quite common. So 6 MSBs will stay above the noise floor while the other 10 hide below it. Suppose one could still hear signals that are below the noise floor one might just be able to squeeze out 8 bits resolution ……
Yes, I have just played around with an old AUNE 1793 DAC ( made in 2008 or earlier perhaps? It sounds crap if compared to my current Audiolab MDAC). Now I coupled it with my friend's home-bake step-up transformers. What a surprise! I would say any capable DIYer should give it a try if you have an old DAC board (easy to do soldering)!
I've obtained very good results also with a fully passive output stage using Sowter 8347. With a typical 20 Ohm I/V resistor across the primary, the output impedance is quite high. Therefore, the input impedance of the following amplification stage MUST be very high. In my case I had about 100 KOhm of a tube amp.
With TDA dac chip, passive conversion with resistor is always better than other method like transformer. I believe Mr T. Lausch prefer the passive resistor than the transformer for that work! Ask him for the better one... Some like John from ECdesigns maid its own wirewound resistor...
i will have a look with some new video chip from philips too for active... said to be better than the AD844 ! But I don't try myself these last ones !
For the output stages, how to say... the war between fet friends and tubes friends is a longer war than the Star Wars saga ! Maybe the secret of tubes is in its band pass width an the distorsion signature in high frequency with a regular attenuation... I don't know.
just my two cents
i will have a look with some new video chip from philips too for active... said to be better than the AD844 ! But I don't try myself these last ones !
For the output stages, how to say... the war between fet friends and tubes friends is a longer war than the Star Wars saga ! Maybe the secret of tubes is in its band pass width an the distorsion signature in high frequency with a regular attenuation... I don't know.
just my two cents
Last edited:
hi sandor, it was your post that drew me to consider step up trafos. however, it's too hard to find one with reasonable cost/performance where i am. for the price of decent trafos i could whip up a tube output or multiple array of discrete i/v circuits.. which is what i plan on doing. will compare lampizator circuit to ecdesign i/v output. more to follow 
since i've already made too many threads on this dac, i'd like to talk about possible tweaks for it on this thread.
got my first 1541 dac running last night, salas shunt everything with 62ohm i/v to fetishizator circuit. souund is already miles better than any 1543, delta sig implementations i have tried.
-i'd like to know what benefit i could hear from 2ma dc offset method,
-and any experience on what ecdesign's i2s attenuation does to the sound?
currently have grundig reclock installed which i feel is a must-do for this dac.
will try waveio's l5 pin header for 44x4 clock signal tonight.
got my first 1541 dac running last night, salas shunt everything with 62ohm i/v to fetishizator circuit. souund is already miles better than any 1543, delta sig implementations i have tried.
-i'd like to know what benefit i could hear from 2ma dc offset method,
-and any experience on what ecdesign's i2s attenuation does to the sound?
currently have grundig reclock installed which i feel is a must-do for this dac.
will try waveio's l5 pin header for 44x4 clock signal tonight.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.