DAC ouput using Transformer

[jkeny]

I'm splitting this off from a very interesting thread that started about a cheap but good DAC board from China & possible ways to mod it. It later got into a very interesting discussion about the use of transformers on voltage out DACs such as the CS4398, or the PCM1793 as used in my Musiland Monitor 01 US DAC. (I have one of the new Vout Sabre DACs ordered so I'll be trying trafos on this & an original Sabre32 also)

I have to thank Bill Fuss for introducing the use of transformers on these DACs and although I know it's not an original idea by him, I can't see much treatment of it on this forum - hence the new thread

Now let me start with the good news (my questions later) - the sound of this arrangement is excellent - natural & dynamic without strain. Compared to the op-amp output stage it's much more life-like. This is my first time to do such an output stage so I'm not speaking as an expert here, by any means, and I still have a lot of experimentation to do but the sound does depend on the trafo used. I tried 2 different trafos - a small cheap encapsulated Talema mains toroidal 16VA 115-0-115:22-0-22 just for my first trial & it sounded good but when I put on my second trafo which is a Dave Slagle made 1:1 nickel based (don't know the %) it showed up the Talema - much more high end clarity & a bit more low end bass as well. These trafos weren't made for this role so they were very much an experiment but have prompted me to ordering a Sescom line output trafo from eBay.

Here's my existing output stage schematic attached - a differential Vout from a PCM1793 into a LPF with gain of 1.83 to bring the level up to 2 Vrms. Instead of OPA2134 my board has an inferior op-amp OP275 but it still sounded good stock. I moved the 1K8 R on each leg down to the cap so as it was now isolated from the DAC output & I could connect a wire to each vacated solder pad on the DAC output side. I will be able to use a switch or jumper to reconnect the original output stage if needed.
(Pic of board is attached before this mod)

I'm running this into an unmodded Lepai Tripath TA2020 amp (which is down for modding shortly)

So now my questions:
- I believe this should work purely differentially into the amp without a ground from the DAC board but just using one side of the trafo differential out as the ground but I get music and lots of hum in this arrangement. Any ideas why?
- I can get it to work beautifully, with using a ground connection to the DAC & one of the trafo differential out wires as the signal, no hum.
- When connected this way I have an unconnected trafo differential out wire - if I put a 100R across this & the ground I get a louder & more rich sound - what's happening here?

[ttan98]

You will this site has lots of useful info., incl. transformers and DAC quality.

http://www.audiodesignguide.com/DAC_final/DacFinal.html

I am interested in the transformers as well, I would be interested in alternative trans. to those from Lundahl which are very expensive.

[Pano]

Welcome to the wonderful world of transformer DACs! I've been doing this a lot and really love it. Most good differential output DACs respond well to this mod. No opamps or LPFs needed.
Yes, the transformer quality matters. I've tested many, they all sound different. High nickel content is good. I have mine custom made by BudP, they are the best.

I know what you mean about the hum. Sometimes I can get away without connecting the low point of the secondary to ground, but mostly not.

Tweaking your load resistor will give different results, but many circuits are not too picky about about it.

You can also use a transformer as the DC blocker on your Tripath input. Just tie the low point of the secondary to the 2.5V supply instead of ground. Works great, no cap needed. Have fun!

[jkeny]

Thanks ttan98 - I did see that site before but I'll read it again in the light of new info

Originally Posted by panomaniac

Welcome to the wonderful world of transformer DACs!

Is the water warm? Speaking of which - did you leave Maui? Why, it's so idyllic

I've been doing this a lot and really love it. Most good differential output DACs respond well to this mod. No opamps or LPFs needed.

One thing - if the DAC chip datasheet states the output min impedance is 1K8 (as the PCM1793 does) should this not be the series R on each leg? And if doing this then putting a cap to ground after this to give a LPF is probably not going to do any harm or is it? It's all new to me & I'm tying to learn the principles/vagaries of this. However, I did build the second channel last night without any LPF and initial impressions are that it sounds the same as the other channel with 1K8+ 1nF LPF (88KHz) - so I'm not sure what gives - the transformers windings measure 200R on each leg?

Yes, the transformer quality matters. I've tested many, they all sound different. High nickel content is good. I have mine custom made by BudP, they are the best.

Can you give any ranking to known transformers that you've used? The thing that usually puts people off trafo use is their cost - I had this Dave Slagle one here which I used but had I only tested it with the cheap Talema, I would not have been so impressed. I'm sure that BudP's are excellent and well worth it (as are Dave Slagle's) but there are some cheaper options for use in this role that I've picked up on like UTC, Sescom, maybe Triad? which seem to offer good bang/buck? Can you give us your experience from your tests?

I know what you mean about the hum. Sometimes I can get away without connecting the low point of the secondary to ground, but mostly not.

So any explanation to this? In my existing config, I have left one of the secondary legs unconnected & just used the other one as the hot lead & ground from the DAC as the feed into the amp.

Tweaking your load resistor will give different results, but many circuits are not too picky about about it.

Is that what I'm doing with the 100R across the above unconnected secondary lead & gnd? What does this do & is there a way to test this to find the optimum. I have a scope (but haven't used it much)!

You can also use a transformer as the DC blocker on your Tripath input. Just tie the low point of the secondary to the 2.5V supply instead of ground. Works great, no cap needed. Have fun!

Cool, I hadn't thought of this one - I had already posted asking about ways to DC connect a TDA1543 to a Tripath but this is good

[jkeny]

One thing that occurs to me on re-reading your post - you mention the low side of the secondary a couple of times, I thought it didn't matter which side of the secondary was thought of as the hot signal lead although by rights to keep phase correct, it should be the high side used. Have I just answered my own question?

On the Tripath DC blocker connection, I'm a bit concerned as I bring the ground from the DAC straight through (bypassing the trafo) & connect it to the low-side of the secondary. If I now tie the biascap pin directly to this low-side which has 2.5V pin it, will this not feed back to the DCA ground and damage the DAC?

[Pano]

Well let me see if I can answer.

Originally Posted by jkeny

- did you leave Maui? Why, it's so idyllic

Yes, Maui no more. I was in the art biz, got hit early and hard by the economic storm.

One thing - if the DAC chip datasheet states the output min impedance is 1K8 (as the PCM1793 does) should this not be the series R on each leg?

No, it's the minimum impedance the DAC wants to work into. That will always be higher than the DCR of the primary. It is a combo of the DCR, inductance and the load reflected back from the secondary. At high frequencies it can rise quite high. With most 1:1 transformers I've measured - about 2.5K across the primary will look like a fairly flat 1K load to the DAC (~10K to 100K on the secondary).

And if doing this then putting a cap to ground after this to give a LPF is probably not going to do any harm or is it?

Prolly not. But I don't think you need it. The transformer will clean up most of it anyway.

Can you give any ranking to known transformers that you've used?

Of the many I've tested I would rate Onetics (BudP) the highest. All the good points of the others with none of the bad. Jensen next. Very nice. Maybe some various Chinese stuff, then Edcor (cheap!). Last for me was Sowter, I just didn't like them. Still need to test Cinemag and maybe Neutrik. But I from what I know about them, I doubt they will best the Jensens.

The thing that usually puts people off trafo use is their cost

Don't forget size and weight. Can be a real pain.

So any explanation to this? In my existing config, I have left one of the secondary legs unconnected & just used the other one as the hot lead & ground from the DAC as the feed into the amp.

I don't understand.

One thing that occurs to me on re-reading your post - you mention the low side of the secondary a couple of times

I just meant the "negative pole." It's all relative.

Hope that helped!

[jkeny]

Originally Posted by panomaniac

Yes, Maui no more. I was in the art biz, got hit early and hard by the economic storm.

Sorry to to hear that

No, it's the minimum impedance the DAC wants to work into. That will always be higher than the DCR of the primary. It is a combo of the DCR, inductance and the load reflected back from the secondary. At high frequencies it can rise quite high. With most 1:1 transformers I've measured - about 2.5K across the primary will look like a fairly flat 1K load to the DAC (~10K to 100K on the secondary).

I measure the resistance across each primary as 200R - it's a dual so 400R in total. Is this the DCR?

Of the many I've tested I would rate Onetics (BudP) the highest. All the good points of the others with none of the bad. Jensen next. Very nice. Maybe some various Chinese stuff, then Edcor (cheap!). Last for me was Sowter, I just didn't like them. Still need to test Cinemag and maybe Neutrik. But I from what I know about them, I doubt they will best the Jensens.

Thanks for this - I'd expect BudP's to be expensive & the Jensens are not cheap either. I bought the Sescom's MI-97 for $50 a pair on a recommendaton - I'll post how they sound when I get them!

I don't understand.

I've attached a crude drawing which I hope explains it. One of my secondary leads is left unconnected (A) to the amp - I just use the other one (B) as the signal into the centre pin of RCA connector & a ground wire runs all the way back to the DAC board. Is this wrong? It seems to be the only way it will work? If I connect both legs without ground I get hum. I discovered that if I connect the A wire across a 100R to ground I get a higher volume (stronger signal) - any explanation for this?

That is why I had pause about the Tripath connection - should I connect the A wire to the Biascap, this will give a 2.5V bias to the B wire but also 2.5V to the ground because of the 100R across it to ground - should I leave out this 100R?

Hope that helped!

It sure did & I'm sorry for all the questions - it's just that transformers are another world to me in this role & there doesn't seem to be much info on here - thanks for your help I really don't know how to calculate the R reflected back from the secondaries & is this R the impedance of the input stage of the amp? Why does my cross-connected 100R increase the volume?

[Pano]

A few answers:

Your circuit will work, I've done it (by accident.) But it's not the best way to do it. And you will often see 2X the output wired like this. You should really connect one end of the secondary to the point where the input cap normally goes. The other end to the 2.5V bias. Think of the 2.5V bias as your ground.

If your transformer has dual windings then you have the ability of wiring them in series or parallel. That will change the ratio of the transformer - and the DCR, as you note. In series is probably best for you. So 400R. With that you have to figure in the inductance of the transfo and the load on the other side. With a 1:1 transformer the reflected load is the same as the secondary load. Thus 1:1

The impedance curve of the transformer will start at DCR at DC (of course) and rise rapidly toward about 20Hz. From there it will rise more with a few kinks and bumps to be a bit higher than the secondary load at 20KHz. As you can imagine, putting a 2K or 3K resistor across this will swamp that rise. You'll get a much flatter curve and your chip will be happier.

Sounds like you got a great deal on your Sescom. The Jensens I used are about $35 each. Still a bargain, I think. Let us know how they sound!

[jkeny]

I'm sorry Mike, I don't understand - can you do a quick sketch please? Also I don't just want to make this specific to a Tripath amp as I may want to plug it into other amps!

Also you say a 2k to 3K across - do you mean across the secondary or across one secondary & ground? See my diagram

Have you heard the Sescoms? How do you rate them?

[WithTarragon]

Replacing the output section with a transformer is a great idea.
See the Behringer 2496DCX threads for a great deal of information that is relevant. It also uses a voltage DAC.

Just a few points. The DAC will need to see some resistance. Without enough load it will try to draw current and this can lead to some problems. Check the manufacturers specs. For instance the AKM used in the Behringer needs to see at least 600-1000 ohm load. This resistor of course can be part of the LPF (RC filter) necessary for the DACs reconstruction filter. If the DAC has a balanced output and you need an unbalanced output, the transformer also makes this an ideal element in the circuit for eliminating the DC that the two signals ride on. BTW, a properly chosen transformer can provide some additional LPF.

The trick is choosing the "right" transformer. They will vary in sound quality and they will vary wildly in price. In terms of price, from high to low the choices can include Lundahl, Sowter, Jensen, Cinemag, Edcor. If you are unfamiliar with transformers, read the tutorial on the Jensen Website.

Good Luck