Buffalo II & transformers

[Vingborg]

Sorry but I don't know why this damn picture upload dosen't work for me - I'm using Flickr and there use to be no problem with that!

I put in the numbers - R1: 195 ohm. R2: 0.000001 ohm. R3 0.5 ohm. Voltage V1: 1.65 Volt.

Are you guys getting any out of this??

Henrik

Hello

All right, I tried to enter the data that I have - dac, load and so on.. The picture look like this:

An externally hosted image should be here but it was not working when we last tested it.

I must admit that I don't understand much of these numbers or the calculater..
I never heard of Thevenin law, but hey, I'm a carpenter

For me it looks like transformer is served an output impedance of 0.49872222 ohm from dac on primary winding when I load the dac with 0.5 ohm!
Is that true??

Can you Marek.. Or somebody else bring a light on this for me

I would be so enormously happy if it is possible for me to present the LL1941 for a source impedance as low as 0.5 ohm!

Very best regards
Henrik

 

[Marek]

Your calculation is ok - with 0,5ohm load resistor you get source of less than 0,5ohm resistance at the expense of output voltage - about 4mV.

 

[Vingborg]

Wow fast answer

In that case there should be no problems by using a Lundahl LL1941 with the ES9018 dac chip. It should provide both very low and high frequencies with no problem at all! Yes?

I wonder why nobody have done this time ago? Load the dac with something below 1 ohm and use high quality step up and a tube or fet gain stage...

Is there any disadvantages by this approach?

Henrik

 

[Vingborg]

I know the output voltage is going to be very low, but that is the only way to make this dac work in current mode, from my point of view.

Yes the TP guys says that it is not possible to run in current mode with a transformer..

I would say that a load resistans of 0.5 ohm is very, very low. And if that isn't.. what is then?

I'm not an electronics engineer so how is it possible to reach a input impedance of 0.5 ohm or even lower without draw the signal down to only a few millivolts?

Best regards
Henrik

 

[Marek]

Wow fast answer

In that case there should be no problems by using a Lundahl LL1941 with the ES9018 dac chip. It should provide both very low and high frequencies with no problem at all! Yes?

Yes.

I wonder why nobody have done this time ago? Load the dac with something below 1 ohm and use high quality step up and a tube or fet gain stage...

How come? This is what Joe Rassmusen proposed many posts ago!

 

[Russ White]

I know the output voltage is going to be very low, but that is the only way to make this dac work in current mode, from my point of view.

Yes the TP guys says that it is not possible to run in current mode with a transformer..

I would say that a load resistans of 0.5 ohm is very, very low. And if that isn't.. what is then?

I'm not an electronics engineer so how is it possible to reach a input impedance of 0.5 ohm or even lower without draw the signal down to only a few millivolts?

Best regards
Henrik

Current/Voltage source is purely a matter of perspective. I have never said any differently To an infinitely large output impedance .5R looks like an ideal load.

I don't have a dog in the hunt. I just happen not to prefer transformers in my signal path - yet that does not mean I could not be convinced to change my mind.

Cheers!
Russ

 

[AR2]

I know the output voltage is going to be very low, but that is the only way to make this dac work in current mode, from my point of view.

Yes the TP guys says that it is not possible to run in current mode with a transformer..

I would say that a load resistans of 0.5 ohm is very, very low. And if that isn't.. what is then?

I'm not an electronics engineer so how is it possible to reach a input impedance of 0.5 ohm or even lower without draw the signal down to only a few millivolts?

Best regards
Henrik

It would be quite important to connect your setup and measure something like 20 - 60 KHz. Transformers are very sensitive to relationship between primary and secondary impedance and load. You will need to observe the curve in order to see if you will need an RC termination. In my experience only after careful measurement and careful selection of load or termination, you will have a flat curve. Without that, you will be amazed how crazy that curve could go.

 

[Vingborg]

Good evening folks

It is really a pleasure to see interest in this thread

How come? This is what Joe Rassmusen proposed many posts ago!

Yes I know.

In post 185/186 which is very interesting posts:

The reason I said: "I wonder why nobody have done this time ago" is because with a step up trafo I can load the dac even lower than I could with a 1:1.. Because loading the dac with 0.5 ohm would require very high gain (52 db around 400 times) to reach say 2 volt.
By using fx 1:16 trafo, the gain after the trafo does not require more than about 25 times and that is what my tube gain stage is providing.
A tube stage with 52 db gain is extremely noisy!
The EF86 tube stage with arround 25 times of gain that I have assembled is sounding sweet and very low noise.

I know something about the step up 1:16 lowers the current 16 times when the voltage is increased 16 times.
A tube stage, especially with EF86 tubes has a very high internal impedance which suits the secondary winding fine!
Again I'm not a engineer so this is just my little cautious theory!

I don't have a dog in the hunt. I just happen not to prefer transformers in my signal path - yet that does not mean I could not be convinced to change my mind.

Cheers!
Russ

Hi Russ

I totally agre with you! I have some serious doubts about all these "copper windings and steel core" in my delicate signal path.
But I really like the sound of electron tubes and I don't know any other way to connect the Buffalo III with my EF86 tube stage by using a transformer! To keep the two circuits appart and do tricks with the loading

I can't afford to by a whole bunch of high performance tranni's.. Just to try them out.. I wish I had - But I suspect these Lundahl LL1931/1941 to be a really good choice for high end sound

It would be quite important to connect your setup and measure something like 20 - 60 KHz. Transformers are very sensitive to relationship between primary and secondary impedance and load. You will need to observe the curve in order to see if you will need an RC termination. In my experience only after careful measurement and careful selection of load or termination, you will have a flat curve. Without that, you will be amazed how crazy that curve could go.

I hope my brother who is a electronic technician, can help me with his scope and so.. And maybe Per Lundahl can tell me something about some network for the transformer..
But.. There is still some test to do with the LL1684 I have borrowed. And with a unknown step up 1:15 I have bought cheaply on Ebay!

I thank You all for your time and patience with me and my ignorant questions

Best from Henrik

 

[Ryssen]

But.. There is still some test to do with the LL1684 I have borrowed. And with a unknown step up 1:15 I have bought cheaply on Ebay!

Do report the outcome...

 

[Joe Rasmussen]

I totally agre[e] with you! I have some serious doubts about all these "copper windings and steel core" in my delicate signal path.

Hi Henrik, I have had to overcome similar 'serious doubts' about transformers in the signal path, but glad I did. Besides, the signal does not travel along those windings like it does in an interconnect or length of wire, cross-coupling (or whatever phrase you want to use) makes it behave entirely different.

Also, how many of us have avoided signal traffos and yet tolerated them in our tube power amplifiers where they are potentially as bad, if not worse, than anywhere else in the signal path. It comes down to a a state of mind.

Transformers are both good and bad. Good one: They can work amazingly well with straight "voltage" DACs summing both phases via 1:1 and generally 2V RMS outs with not additional electronics, no power supplies etc. This is because "voltage" DACs generally have low output impedance, a fraction of that of the Sabre DAC. So I would not personally use transformers at all with the Sabre DAC. But if I did, then I would use something like the schematic you re-posted as I would then like to use tubes.

Cheers, Joe

.

 

[Vingborg]

Do report the outcome...

I promise, but things aren't going that fast in this house.. I have two small kids
But off course I will soon test a little!

Hi Henrik, I have had to overcome similar 'serious doubts' about transformers in the signal path, but glad I did. Besides, the signal does not travel along those windings like it does in an interconnect or length of wire, cross-coupling (or whatever phrase you want to use) makes it behave entirely different.

Also, how many of us have avoided signal traffos and yet tolerated them in our tube power amplifiers where they are potentially as bad, if not worse, than anywhere else in the signal path. It comes down to a a state of mind.

Transformers are both good and bad. Good one: They can work amazingly well with straight "voltage" DACs summing both phases via 1:1 and generally 2V RMS outs with not additional electronics, no power supplies etc. This is because "voltage" DACs generally have low output impedance, a fraction of that of the Sabre DAC. So I would not personally use transformers at all with the Sabre DAC. But if I did, then I would use something like the schematic you re-posted as I would then like to use tubes.

Cheers, Joe

.

Hi Joe
Thank you for joining in.. Do you know that you are the main reason for my big interest in this whole transformer matter! Your pleasant way to explain things suits me just fine

Hmm, I had the feeling that you where using Sabre 9018 with transformers and tubes in your setup.. I did make a research on your website but I could not see much about your ref. dac.

I realize this thread is about ES9018 and transformers, gradually also tubes.. But now I'm curious - If you are running with the ES9018 dac, how did you connect it then? I mean: Is it a finished product, some diy solid state output stage, opa..???

When it comes to all, I think I'll end up with the LL1941 with my pentode gain stage.. Even though I have never listened to LL1941. But I have listened to LL1684 with my pentode gain stage with standart cheap components, and it sounded absolutely fantastic!!

I have the feeling that the 1941 is the same or better quality as 1684 + I can load the dac "harder" with LL1941

Audio Note, they use transformers for I/V conv. in their dac's and I do not agree in the way AN build some of the circuits especially their psu, but still the Dac 3/4/5 with AD1865 chip, transformers and tubes is very high regarded

Best from Henrik

 

[CFT]

Perhaps this opa660 config can replace the transformer?
(input stages identical in the OTA and buffer sections on chip)

Say make the input loads one ohm or less...

Does this chip have enough bandwidth to cope with the 9018?

 

[Marek]

We are playing with transformers not because we don't know OPAs...

 

[Magz]

I don't buy this theory that the 9018 in voltage mode doesn't work well with a transformer. I'm doing it and it sounds great! No limitations at either end of the spectrum. You just need to make sure that the transformer has high primary inductance - the LL1676 (200H primary inductance according to my LCR meter) works very well in this regard. 200R is not much source impedance at all - you can run a 6SN7 into 200H and get good full range response and that tube has a plate resistance of 7000R!

In my case the LL1676 separates the different DC offsets from the DAC and the 26 tube grid, and gives me a 2:1 stepdown so that I don't overload the AVC on the output of the 26. I can even get the Rasmussen rolloff without needing a cap. Give it a shot!

 

[Joe Rasmussen]

I don't buy this theory that the 9018 in voltage mode doesn't work well with a transformer...

Oh indeed, it does work, but just not as well - I tried for nearly a year to get it to work as well as possible and it just comes down to the fact that transformers likes to be driven by a low source impedance and that makes a noticeable difference. I know somebody who drives transformers with all eight phases of the ES9018 in parallel (mono and hence he has two DACs) and gets under 100 Ohm and that would make quite a difference from 4 phases, 195 Ohm - and he likes it. But a "voltage" DAC typically has 20 Ohm output Z and that's where I would like to see it.

In my case the LL1676 separates the different DC offsets from the DAC and the 26 tube grid, and gives me a 2:1 stepdown so that I don't overload the AVC on the output of the 26. I can even get the Rasmussen rolloff without needing a cap. Give it a shot!

Look, if that sounds good, then congrats. But re the "roll-off" that Ken Newton dubbed the "Rasmussen Effect" and got s number of people's noses out of joint, with transformers will most likely need an "over-cooked" Zobel network.

Ask the question, does it have a rising response at 20KHz (relative to 1KHz)? If so, try the following:

Try a 15nF film cap in series with 1K resistor as a Zobel network across the secondary. Now measure if the response is near or just short of -2dB @ 20KHz. Adjust resistor value to get that result. Now check the response at 10KHz (relative to 1KHz) and check that it is about -0.5dB there. If the response is more that -0.5dB to -0.6dB area, then the cap value is too large (or vice versa), reduce to 10nF (or increase capo value if response is not far enough down at 10KHz). Now readjust R value gain to near -2dB at 20KHz, then again check 10KHz and continue in this vein till you get the right result. Then use the same final values on both channels. So the final response is typically -0.5dB @ 10KHz and near or almost -2dB @ 20KHz.

Clear as mud? It's worth the effort, so please try it and report back.

I use a Denon test disk that has the three required sine wave tracks, 1KHz, 10KHz and 20KHz. There is also software that can produce the files that can be played or burnt onto a CD-R, so that way all bases are covered, disk or files.

Cheers, Joe

.

 

[Magz]

Oh indeed, it does work, but just not as well - I tried for nearly a year to get it to work as well as possible and it just comes down to the fact that transformers likes to be driven by a low source impedance and that makes a noticeable difference. I know somebody who drives transformers with all eight phases of the ES9018 in parallel (mono and hence he has two DACs) and gets under 100 Ohm and that would make quite a difference from 4 phases, 195 Ohm - and he likes it. But a "voltage" DAC typically has 20 Ohm output Z and that's where I would like to see it.



Look, if that sounds good, then congrats. But re the "roll-off" that Ken Newton dubbed the "Rasmussen Effect" and got s number of people's noses out of joint, with transformers will most likely need an "over-cooked" Zobel network.

Ask the question, does it have a rising response at 20KHz (relative to 1KHz)? If so, try the following:

Try a 15nF film cap in series with 1K resistor as a Zobel network across the secondary. Now measure if the response is near or just short of -2dB @ 20KHz. Adjust resistor value to get that result. Now check the response at 10KHz (relative to 1KHz) and check that it is about -0.5dB there. If the response is more that -0.5dB to -0.6dB area, then the cap value is too large (or vice versa), reduce to 10nF (or increase capo value if response is not far enough down at 10KHz). Now readjust R value gain to near -2dB at 20KHz, then again check 10KHz and continue in this vein till you get the right result. Then use the same final values on both channels. So the final response is typically -0.5dB @ 10KHz and near or almost -2dB @ 20KHz.

Clear as mud? It's worth the effort, so please try it and report back.

I use a Denon test disk that has the three required sine wave tracks, 1KHz, 10KHz and 20KHz. There is also software that can produce the files that can be played or burnt onto a CD-R, so that way all bases are covered, disk or files.

Cheers, Joe

.

Nah, I didn't have to adjust the Zobel at all to get the rolloff. All I had to do was adjust the primary DC resistance of the LL1676.

I set up the Zobel to give me a nice square wave at 2kHz and 10kHz and left it that way (1k, 3nF). I placed a 5k pot between the two series-connected primary windings of the LL1676 and adjusted it to get the proper rolloff at 20kHz vs 1kHz - it was fast and easy, just turning the dial until the rolloff was as desired on my scope. Then I replaced the pot with a TX2575 metal foil resistor of the exact resistance needed, which happened to be 3k3 (Texas Components will custom make any value you need up to 100k). Voila!

I tried a few different values of the rolloff and liked -1.3dB the best, so that's where it is now. Lots of air, clean punchy bass, smooth midrange.

 

[Vingborg]

Hello folks

We are playing with transformers not because we don't know OPAs...

I've got your point, but if transformers and gain stage after isn't the "right and most suitible" way of using the ES9018 chip, then what is then? Is the Legato and Opus stage from TP really the only circuits who can make this dac do the best performance ever? At least on paper, because how it all sounds is up to the individual to judge
Therefore, I mentioned fet and op-amps
I don't know much about fet and opa.. Still it could be interesting to see schematics from a McIntosh MCD500 cd-player, it use ES9018 chip and was said to be one of the best sounding digital players! Or maybe the Gryphon Kalliope dac!! Yes sir!

Anyway.. Yesterday I powered up my dac prototype. Buffalo III loaded with 1,5 ohm from each phase to ground. Going into LL1684 1:1 Lundahl tranni's and futher into high gain (150 times and plenty noise) pentode EF86 tube stage. Sound was very good, really! but... It was not better than my first experiment with dac into LL1684 1:1 with no resistor to ground! And futher into low gain EF86 tube stage (25 times and low noise)
I could not hear any difference between loading the dac with 1,5 ohm and high gain, or no load and low gain - Maybe that's just me....

The next days I will try out these brand-unknown step up 1:15 tranni's that I got, load the dac and use the 25 times EF86 gain stage and see how it turns out

I use a Lundahl LL1676 with my Buff III, wired 2+2:1+1 for a 2:1 stepdown, no ground connection for the output of the Buffalo, just straight into the two sides of the primary. In this configuration it has more than 200H primary inductance, which is more than enough to provide thunderous bass performance with the ES9018.

The Lundahl provides HF filtering and I've arranged the secondary snubber plus primary resistance to give me the "Rasmussen rolloff" of 1.3dB at 20kHz. Sounds EXCELLENT feeding the output stage, a 26 DHT tube in parafeed to a Bent Audio silver autoformer!

I am so satisfied that I have no intentions of tweaking this any further. It destroys the PCM1704 DAC in my CD player. Not even close.

This is very interesting!
Provide the Buffalo with a step down trafo with extremely high primary inductance and a low gain tube stage!
I don't know if any of you guys can follow me. I'm thinking of these small value termination resistors to ground - They kill the music signal to allmost nothing..
By letting the dac operate with no resistors into a step down trafo with high primary inductance. And - Maybe I'm crazy but my "unknown" step up 1:15 tranni's: Turn them around to step down 15:1. The secundary will now be the primary with 18H inductance. The signal after the 25 times gain stage will surely be too high so resistors from dac to ground could be the solution for this, but the value of these will definitely not be as small 0.5!
Is it mindcrap or..?

I'm not using any load resistors letting ES9018 to work with "voltage mode" because I really can't see any real advantages to work with "current mode" except some better -dB numbers which are negligible to me.

If money was no object I would order a custom supermalloy trafo from SAC Thailand with SILK Nano winding technology SAC Thailand

Marek

Now that is a kind of my point and what I am afraid that I experience with these resistors!

Best regards
Henrik

 

[AndriyOL]

Is the Legato and Opus stage from TP really the only circuits who can make this dac do the best performance ever? At least on paper, because how it all sounds is up to the individual to judge
Therefore, I mentioned fet and op-amps

My mistake about your preferences, sorry!)

As I've read in the net the best choice for ES9018 could be discrete fet I\V stage, besides transformers. The most prefarable for me would be Erno Borbely all fet I\V converters.
Has anyone seen those schematics?

 

[Magz]

Hello folks


This is very interesting!
Provide the Buffalo with a step down trafo with extremely high primary inductance and a low gain tube stage!

Best regards
Henrik

It sounds very, very good. No worries about "high" output impedance of the ES9018 when you have 200H to run it into. Hell, I even added 3k3 MORE resistance to the primary!

All I can say is give it a try. You'll be surprised how good it sounds.


PS: My DAC is a Buffalo IIIse; I noticed I didn't add the "se" to my previous post so just to clarify...

 

[CFT]

Hi Marek, have you tried the ~one ohm loaded current mode yet with your LL1931?