Thanks qusb for sharing your experience on this matter - I was not able to see in the Fischer catalogue what temperature it was rated at - but anyway got your point and will probably look for a better rated heat sink - just to be on the safe side
Thanks
Rolle
Just as an aside - I've started discussions with a custom heat pipe design firm. I'm expecting that if I pursue this, it'll result in a custom case. I have the NTD1 in mind, but the design should accommodate components with similar heat dissipation requirements. The basic design will be very similar the Owen's custom case (bottom plate, sinks on the side), with the bottom plate having embedded heat pipes, and the side sinks/radiators dumping the heat. The bottom plate will be marked such that it can be drilled/tapped without getting into the heat pipes.
Anyway, I'm early in talks and it may not come about (cost, etc.). I mention it to gauge any interest, as it's alot of effort for one or two cases. There are cases out there that will work (we're still knocking around one such case in this thread.), but I'm intrigued by the thought of a more compact case that can handle the thermal requirements of this awesome IV.
So it's premature to start any sort of list, but if this is something you'd want - let me know.
Anyway, I'm early in talks and it may not come about (cost, etc.). I mention it to gauge any interest, as it's alot of effort for one or two cases. There are cases out there that will work (we're still knocking around one such case in this thread.), but I'm intrigued by the thought of a more compact case that can handle the thermal requirements of this awesome IV.
So it's premature to start any sort of list, but if this is something you'd want - let me know.
yeah I looked at and still am looking at the flat cool pipes but I wouldnt want a dac case that looks just like a poweramp if getting something totally custom (as it looks like I am) the heatsink will be milled and not look like a normal finned heatsink. 100W is a lot for a dac, but its not really a lot on the greater scheme of things.
Finally got to see the board 
And two lamish questions cropped up. Apologies if answers were already provided.
Is there a digikey p/n for the regulator sinks? None of what i have at hand seems to match the footprint well.
How do people mount the output caps? Offboard? None of the caps i have could possibly fit. Questionable if the bypasses for the caps will fit alone Holes too small as well...
As i'm a bit reluctant to solder Wima MKPs i am considering an amorphous Lundahl instead. Has anyone explored a similar route?
And two lamish questions cropped up. Apologies if answers were already provided.
Is there a digikey p/n for the regulator sinks? None of what i have at hand seems to match the footprint well.
How do people mount the output caps? Offboard? None of the caps i have could possibly fit. Questionable if the bypasses for the caps will fit alone
Holes too small as well... As i'm a bit reluctant to solder Wima MKPs i am considering an amorphous Lundahl instead. Has anyone explored a similar route?
in one foul swoop you would decrease bandwidth and distortion performance with the TX, plus finding some that dont mind +/-45VDC across them without saturating.... not always an easy task
the holes are fine for any solid core lead ive got. both of the auricap options will fit the board, the stranded lead option of the auricaps leads may be too thick to fit through because they have 2 paralel stranded leads twisted together to match the 2 parallel caps inside, but are easily soldered on top, the radial lead one has solid core leads and fits. my vcap lead fits, my mundorf cap lead fits...
but ive been using offboard 'clusters' of auricap||VCAP CuTF near the XLRs up till now because I have another platform full of dac stuff that stacks on top so even if the caps fit they would make the whole stack too high and the dac input wiring longer than I would like. offboard is fine for the caps, perhaps the large DC bias on the wiring may even help to keep noise low =) but the IV conversion has already happened and the output is driven by a chunky power mosfet, so I really dont think its any problem just mounting them near the XLR.
if you are mounting off board, just remember not to install the 100K resistors to ground at the output pads and install them across the XLR instead (pin 2->1 and 3->1 and ground pin 1 to chassis ground. this is important to avoid/drain the DC transient you get before the caps are charged.
the heatsinks are just pretty standard EA-T220-38E but you could increase that height a touch.
the holes are fine for any solid core lead ive got. both of the auricap options will fit the board, the stranded lead option of the auricaps leads may be too thick to fit through because they have 2 paralel stranded leads twisted together to match the 2 parallel caps inside, but are easily soldered on top, the radial lead one has solid core leads and fits. my vcap lead fits, my mundorf cap lead fits...
but ive been using offboard 'clusters' of auricap||VCAP CuTF near the XLRs up till now because I have another platform full of dac stuff that stacks on top so even if the caps fit they would make the whole stack too high and the dac input wiring longer than I would like. offboard is fine for the caps, perhaps the large DC bias on the wiring may even help to keep noise low =) but the IV conversion has already happened and the output is driven by a chunky power mosfet, so I really dont think its any problem just mounting them near the XLR.
if you are mounting off board, just remember not to install the 100K resistors to ground at the output pads and install them across the XLR instead (pin 2->1 and 3->1 and ground pin 1 to chassis ground. this is important to avoid/drain the DC transient you get before the caps are charged.
the heatsinks are just pretty standard EA-T220-38E but you could increase that height a touch.
Last edited:
Let me guess, transformers are not your favourite coupling devices
Still, the voltage issue you mention is generally a non-issue - there is no dc potential across the windings as the two drains are at the same potential. Not sure if anyone can say exactly by how much this will deviate at different temperatures but i don't anticipate any issues. As for bandwidth and distortion, 80kHz is more that fine with me and i don't really see much point reducing thd below -80db.
Will certainly try it both ways, just have to find some space for the caps.
And thanks for the heatsinks specs.
then forget the IV stage and just connect the transformer to the output directly.... the whole point of this IV stage is to get great performance with only a handful of parts, great length in dissipation, bulk, power supply, heat are gone to, to gain this performance
now do you see my point of view?
actually I have used TX with ESS before (Bud Purvine's Onetics), happily for a while, but I did find they dulled dynamics some, effortless, but too 'smooth' for my liking.
you will still get DC across the windings, they are unlikely to be that symmetrical, even if you match everything on the D1, because they reflect the impedance seen on the other side, which is going to be unbalanced
Last edited:
Let me guess, transformers are not your favourite coupling devices
Still, the voltage issue you mention is generally a non-issue - there is no dc potential across the windings as the two drains are at the same potential. Not sure if anyone can say exactly by how much this will deviate at different temperatures but i don't anticipate any issues. As for bandwidth and distortion, 80kHz is more that fine with me and i don't really see much point reducing thd below -80db.
Will certainly try it both ways, just have to find some space for the caps.
And thanks for the heatsinks specs.
That's correct, there is no DC across the +/- terminals, and the net DC to GND of each leg is roughly 23V IIRC.
You will see a few mV depending on how accurately you set the voltages, and it will vary by a few mV over temp. If you have matched transistors, and you set everything carefully, you might be able to get away with it sans DC blocking capacitor.
Just for the record though, I fully agree with qusp... I don't know why anyone would ever opt to use a transformer when a capacitor is an option. There are applications where transformers work just fine, but in this case you will see a significant reduction in BW, some phase anomalies, much higher THD+N, and a possibly some issues maintaining low enough DC across the outputs.
I'm working on a fully DC coupled variant right now, but there are some issues with turn on and turn off transients, and you have to live with an op-amp in the circuit
Cheers,
Owen
I gathered thats the only format he is considering.
I mentioned the unbalanced reflected impedance on the other side of the TX, most amps have pretty unbalanced impedance +/- inputs, I would expect this imbalance to be reflected in the DC across the windings? thats a question, maybe there is something i'm not considering. this would be the case no matter how perfectly you have matched the D1
youre right though, I wasnt thinking re the actual voltage not being the full rail voltage, rather than just the 23v
doh, dont try to be clever at midnight
Last edited:
there are some issues with turn on and turn off transients, and you have to live with an op-amp in the circuit
Transients are unavoidable in a dc circuit. If a parallel/shorting muting relay is allowable this shouldn't present any problem. Is the opamp only a servo?
I am not dead set on transformers. In fact there are very few transformers whose sound i like. Among other things they provide ground isolation and filtering - worthwhile properties in a dac plus BAL-SE_BAL conversion, which in this particular case seems quite handy.
Don't get me wrong... I fully encourage some experimentation with transformers, I'm just making it clear that they are not my cup of tea.
There are two more specific applications I can think of where I actually would encourage the use of a transformer:
1. If you need more output voltage but don't want to change the circuit.
2. If you must have SE output.
Number 1 can also be accomplished by increasing the resistor values, but that will reduce idle current and impact THD+N. If you needed 4V out then increasing resistors is probably the better option, but if you need more like 8V out, then you'd probably be best with a transformer.
Number 2 can be accomplished with an op-amp (which is what I would do), but I can understand someone being drawn to the simplicity and passive nature of the a transformer to accomplish the same thing.
When the application only requires DC blocking like it does with the NTD1, then I find it hard to make a good argument for transformers. As soon as you throw impedance conversion or BAL/SE conversion into the mix, then the argument for using a transformer becomes a lot stronger.
Cheers,
Owen
There are two more specific applications I can think of where I actually would encourage the use of a transformer:
1. If you need more output voltage but don't want to change the circuit.
2. If you must have SE output.
Number 1 can also be accomplished by increasing the resistor values, but that will reduce idle current and impact THD+N. If you needed 4V out then increasing resistors is probably the better option, but if you need more like 8V out, then you'd probably be best with a transformer.
Number 2 can be accomplished with an op-amp (which is what I would do), but I can understand someone being drawn to the simplicity and passive nature of the a transformer to accomplish the same thing.
When the application only requires DC blocking like it does with the NTD1, then I find it hard to make a good argument for transformers. As soon as you throw impedance conversion or BAL/SE conversion into the mix, then the argument for using a transformer becomes a lot stronger.
Cheers,
Owen
but why would you put it after the NTD1? if you were going to use a TX and -80db THD is good enough, you should just connect the TX directly to the DAC, by putting the tx after the D1 you have just thrown away all the goodness you have gained in the first place with great effort/expense and chassis realestate
in the case of direct IV with TX vs NTD1 with caps its just a matter of taste, but putting it after the D1 just doesnt compute for me i'm afraid. you get all of the benefits mentioned above by connecting directly, but a lot cheaper and in a lot smaller space.
in the case of direct IV with TX vs NTD1 with caps its just a matter of taste, but putting it after the D1 just doesnt compute for me i'm afraid. you get all of the benefits mentioned above by connecting directly, but a lot cheaper and in a lot smaller space.
Last edited:
wouldnt you get better performance just by using half the D1 output? normally I wouldnt say that, but with transformers you are going to lose a lot more than 3db
anyway good luck with that analogue_sa sorry for my fervor here, the main appeal TX ever had for me was the simplicity of just connecting directly with a transformer-> IV resistor maybe a simple jfet buffer after it. but in the end I just didnt like it
Last edited:
The main reason, and it's a big one, is operating the DAC in current mode rather than voltage mode. If you connect a transformer directly to the output of the DAC you'll be running in voltage mode which, quite frankly, sounds like crap.
With the NTD1 transformer coupled on the output, at least you're running the DAC in current mode into the NTD1. You're also driving the transformer from a low impedance source which is better than driving it off the DAC directly.
As for the SE output thing, you really do give up a lot deriving the signal from only one leg of the D1. THD+N increases above -100dB and I actually don't much like the way it sounds either. You're much better off feeding the differential output to an op-amp to get to SE, or as I mentioned above, running a BAL-SE transformer. At least that way you get some of the differential cancellation advantages which impacts both noise and distortion.
Again, I'm not arguing with you overall... I would exhaust pretty much every other option before going to a transformer, but not everyone feels that way. Where I tend to strongly disagree with the use of transformers is when people resort to using them simply out of an irrational fear of having a capacitor in the circuit. People tend to associate negative things with series caps, and positive things with transformers, when it's very much the opposite. In a blind AB test, I cannot detect a good capacitor in series with the signal chain, but I can definitely detect a transformer.
Cheers,
Owen
hmm but weve established that hes happy with -80dB and we both agree that thats probably a good thing because it wont be too far wrong
. do you think you could still pick ESS current mode and ESS voltage mode, or ESS single phase mode, once its been reduced to -80THD?that was my point, there is a pretty severe leveler in the circuit that dominates all the other factors
as for reality of what I would consider or use myself we are totally on the same page. but if using a device that knocks the THD down that low, burning 50W and taking up more than 10x the space to get the THD to burn (most of the reason to take those extreme steps in the first place).. thats nuts IMO
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.