Tube line stage output impedance

[kevinkr]

Yes, it's news. Don't virtual ground I/V converters offer something in the region of milliohms of impedance?

Yes I'm confused too - I've used resistors down to a couple of ohms with no problems ever, (pcm63, AD1860 and AD1862) and my current THS4131 based i/v set up used with PCM1798 is certainly in the milli-ohm range from dc to at least a hundred kHz. (Currently I'm using these PCM1798 dacs in mono mode, current summed into THS4131s.) My understanding is that the chip generates 4mA (1798) or 8mA (1794) per channel at full scale regardless of whether it is a virtual earth synthesized by a transimpedance amplifier or a dead short to earth. (The chip really cannot tell the difference.) Note that a short to a supply rail could be another matter.

Revintage I'm just wondering how your experience differed to make you come to a different conclusion..

 

[kevinkr]

I've been studying up on this 5842 tube.

I could use it to amplify a very low output from my DAC. The advantage is the lower passive I/V resistor one uses, the more linear the DAC chip behaves.

Would it be insane to feed a 5842 a 0.025V rms signal from the DAC? This would give me 1 Vrms output.

My concern would be low level detail (-96 dB down) would be feeding the 5842 in the sub-MicroVolt level.

Of course most masterings don't use the least significant bits, but would I be amplifying more noise than music? Or is this an accepted practice with the 5842?

Yes I did something quite similar in one of my old dacs, where I noted some improvement with values down to 10 ohms, the trade off being increased noise. The 5842 has a very low internal noise resistance so you would probably be ok doing this. You can always change to a slightly higher value of resistance if need be.

I would not get too concerned over those last few LSB getting lost in the noise particularly in the 24 bit realm as nothing else in your system has comparable dynamic range - the rest of the amplification chain contributes noise and then there is both the question of your dynamic range and the background noise levels in your listening room. In a real sense I would doubt any existing hifi system from electrical input to speaker acoustical output has even close to a full 16 bits of dynamic range. Practically speaking with less than the theoretical 96dB possible, an extreme example assuming an ambient noise level of just 30dBspl (that's pretty quiet) to achieve the full 96dB electrical dnr the maximum acoustical output of the system would have to be 126dBspl.. (ouch)

Now I'm not saying that you should not be concerned about noise, but most program material has a dnr of much less than 60dB so in theory (reality) you are never going to see voltages in the realm of concern if you chose resistor values judiciously.

 

[revintage]

Sorry, my way of express myself in English isn´t always the best . My deepest regrets for that.

What I ment was that you won´t get any signal left with an extremely small resistor loading, 0 ohm being the smallest possible.

Passive resistor-loading was the subject, not OP-amp I/Vs.

If we use a resistor "in the milliohm range" like 10mohm and for example have a DAC with 5mA current output (not "Imax") we would get an output signal of 0,05mV. Hope my math is better than my english, .

So kevinr, your remark of acceptable results down to 10ohm shows where the practical limit due to noise is.

I have found that transformer loading in combination with resistor loading resulting in a load in the ballpark of ca 50 ohm works best for me. But this is just a matter of taste

 

[pietjers]

Revintage, thank you for the very interesting choke loaded mu follower schematic!!

For several years I am using standard mu followers (with 5687 and 6N30pi double triodes), and in my system these circuits really sing!

The theoretical advantage of a choke over a fixed resistor between the upper and lower tube is clear, but could anybody share his experiences regarding the sonic results of the standard mu follower vs the choke loaded version? How do they compare?

Does anybody know links with more information regarding choke loaded mu followers?

Thanks in advance.

Peter

 

[regal]

I have found that transformer loading in combination with resistor loading resulting in a load in the ballpark of ca 50 ohm works best for me. But this is just a matter of taste [/B]

Which step-up transformer are you using? Is there a loss off transperancy? Probably no worse than a filter inductor?

 

[revintage]

I earlier used a Lundahl LL1674 connected 1:4 with secondary termination. You can find info at K&K. In this combination the filter works as a lowpass filter.

Nowadays I use a Sony player with a voltage DAC feeding a passive balanced LCR filter to a custom quadfilar from Dave Slagle at Intactaudio. The filter is needed due to the extreme bandwith of the quadfilar.

 

[Allen Wright]

Would it be insane to feed a 5842 a 0.025V rms signal from the DAC? This would give me 1 Vrms output<<

No, 5842's (WE417) are often quiet enough for a MC cartridge input stage, which are typically 100x lower in output level (0.25mV)

>>My concern would be low level detail (-96 dB down) would be feeding the 5842 in the sub-MicroVolt level<<

No problem.

But I wouldn't use a 5842 as they can be HORRIBLY microphonic, and oscillate without warning unless you take extreme measures. An even better tube is the Russki 6C45, much quieter and far less microphonic, yet similar gM and gain. And way less expensive!

Good hunting, Allen

 

[regal]

yea I considered the 6C45 but they are often difficult to match as there is a lot of variation between them, the 5842 is actually cheaper now days.

 

[Allen Wright]

Good luck, then.

Regards, Allen