Re
Sure, simulation a great tools because you save a lot of time experiencing on the screen what would take ages to build and measure. As it does not model all the parameters of real life, there is a stage where we have to build it to know. Still, if sim says 0.000V+/-5µV I assume it will stay within 25mV around 0V
It is coupled with a 4.7µF film in real life
True, here are real measurements :
Oh well, I was talking about I/V conversion using Op-Amp. The super common grid has the same output impedance (2.2kΩ which is the plate resistance).
Enjoy !
greg
Hi,
I would not assume so. You can find quite a few 100mV bias variation with real tubes...
Mine is coupled with 2.2uF Tinfoil & Film and 100nF Silver Mica...
But I was NOT referring to the coupling cap, but to the Cathode/Source bypass and coupling cap internal to your feedback loop.
I have seen worse, but you seem to have lot of spuria in these FFT's and you cut the FFT off at 5KHz.
Attached one of mine, TDA1541, passive I/V and open loop tube gainstage at 0dbfs, this is without oversampling or other digital filtering, 44.1KHz sample rate.
It can be seen there is "much" 2nd HD (about as much as most speaker drivers produce at 1 Watt input power), but all higher harmonics are < -90dB. The 2nd HD falls quite quickly with lowering level...
Sorry, I misunderstood and thought you where talking about tube circuits. There is no need to buffer a correctly applied I/V OP-Amp.
Ciao T
Sure, simulation a great tools because you save a lot of time experiencing on the screen what would take ages to build and measure. As it does not model all the parameters of real life, there is a stage where we have to build it to know. Still, if sim says 0.000V+/-5µV I assume it will stay within 25mV around 0V
I would not assume so. You can find quite a few 100mV bias variation with real tubes...
It is coupled with a 4.7µF film in real life
Mine is coupled with 2.2uF Tinfoil & Film and 100nF Silver Mica...
But I was NOT referring to the coupling cap, but to the Cathode/Source bypass and coupling cap internal to your feedback loop.
I have seen worse, but you seem to have lot of spuria in these FFT's and you cut the FFT off at 5KHz.
Attached one of mine, TDA1541, passive I/V and open loop tube gainstage at 0dbfs, this is without oversampling or other digital filtering, 44.1KHz sample rate.
It can be seen there is "much" 2nd HD (about as much as most speaker drivers produce at 1 Watt input power), but all higher harmonics are < -90dB. The 2nd HD falls quite quickly with lowering level...
Sorry, I misunderstood and thought you where talking about tube circuits. There is no need to buffer a correctly applied I/V OP-Amp.
Ciao T
Attachments
Wow! euro21, that you so much
I have downloaded LTSpice and will be experimenting with it. Can you explain what does the SINE (1.5mA 1.5mA 1k) means at the bottom of the screen shot. My guess is 1k sine wave from 1.5mA current source?
This is only the simulation.
The current source represents 1541a, 1.5mA offset, and 1.5mA amplitude sine (1KHz).
Attachments
re re
Well, the schematic for the TDA1541 use an OpAmp to compare the cathode potential with the ground. If the cahode moves, the OpAmp correct as fast as it can the operating point of the tube by acting on its grid. A CRC filter ensures the OpAmp reacts for variations at lesser than 10Hz and does not mix with the mosfet's feedback. After that, I agree OpAmp are not perfects but I can't believe it would let an offset greater than 0,1mV on the cathode.
Yes me too
Yes, I wanted to pay more attention to H1 to H5 harmonics. Higher harmonics are melted in the background noise.
That's interesting to see how an open-loop circuit produces high level of H2 whereas high feedback schematic will cancel them and only produce low levels of H3 and H5.
[/QUOTE]
Enjoy !
greg
Hi,
This is what is called a "servo". With Servo's I usually shoot for an 0.1Hz cutoff, not 10Hz and if I can I prefer to use a 2nd order loop.
Servo's can indeed correct the DC operating point, however if they are not very well optimised and implemented I find the worse than a straightforward coupling cap, even an electrolytic one at that.
Yet that is not all a feedback circuit does. In principle an open loop circuit can be made to have lower distortion than what I show (0.01% is not really difficult), I cannot see the point at all. We can design with feedback for as many zeros as we like, yet this says nothing about "good sound"...
Ciao T
This is what is called a "servo". With Servo's I usually shoot for an 0.1Hz cutoff, not 10Hz and if I can I prefer to use a 2nd order loop.
Servo's can indeed correct the DC operating point, however if they are not very well optimised and implemented I find the worse than a straightforward coupling cap, even an electrolytic one at that.
Yet that is not all a feedback circuit does. In principle an open loop circuit can be made to have lower distortion than what I show (0.01% is not really difficult), I cannot see the point at all. We can design with feedback for as many zeros as we like, yet this says nothing about "good sound"...
Ciao T
Hi,
The TDA1541 is rated as having a -4mA (peak-peak) output nominal, with digital silence hence at -2mA. The tolerance range is given as 3.4 to 4.6 mA, however in my experience the spread is considerably smaller in reality.
To model the TDA1541 use a -2mA offset and a 2mA peak (1.4mA RMS) AC current source.
Given that my circuit operates the tube essentially with nearly no bias (-0.1V nominal) one must look at 0V Grid curves for a given tube. My data sheets for the 6N30 show around 70mA anode current at 70V and 0V on the grid. I doubt this circuit can work well outside a simulator, with over 5W Anode dissipation per section.
By comparison the 6DJ8/ECC88/6922 has around 16mA at 70V and 0V Grid, which is a bit over 1W per section.
Ciao T
The TDA1541 is rated as having a -4mA (peak-peak) output nominal, with digital silence hence at -2mA. The tolerance range is given as 3.4 to 4.6 mA, however in my experience the spread is considerably smaller in reality.
To model the TDA1541 use a -2mA offset and a 2mA peak (1.4mA RMS) AC current source.
Given that my circuit operates the tube essentially with nearly no bias (-0.1V nominal) one must look at 0V Grid curves for a given tube. My data sheets for the 6N30 show around 70mA anode current at 70V and 0V on the grid. I doubt this circuit can work well outside a simulator, with over 5W Anode dissipation per section.
By comparison the 6DJ8/ECC88/6922 has around 16mA at 70V and 0V Grid, which is a bit over 1W per section.
Ciao T
Thanks Thorsten, I think you are referring to The Universal Tube Output Stage at The Universal Tube Output Stage-- State-of-the-art Performance Made Easy | Diy HiFi Supply ,well I have posted the link.
Little overdissipated tube.
Anode dissipation limit is 4W/sections. Fool-proof safe range up to 80-85%.
English version:
Glass Tube 6N30PDR
Attachments
Ok! I went through TDA1541A datasheet again and this time managed to find on the end of Page 7, "To ensure no performance losses, permitted output voltage compliance is ±25 mV maximum". Thus the maximum size of resistor becomes 17 Ohm.
Hi,
This is wrong.
First, the Philips Datasheet is conservative. At AMR we did run extensive tests on that. The TDA1541 can handle more but not a LOT more. 50 Ohm I/V resistance is fine.
Secondly, if you want to stay within the 25mV compliance reliably for all possible output currents of the TDA1541 range you need 5.4 Ohm I/V conversion resistor (0.025V / 0.0046mA = 5.435Ohm).
My original "Adagio" DAC with TDA1541 used approximatly this value of resistance and a 1:15 Step Up Transformer. It also used an SRPP output stage tuned for lowest distortion, due to concerns about some of the ultrasonic "images" produced by the absence of oversampling folding back into the audio range.
While sounding good in the end we found the sound better without transformer and with a circuit that did not use agressive cancellation of even order harmonics. This then brings us back to the 2005 TDA1541 Analog stage.
For fun you could try using the E180F triode connected and with the I/V circuit scaled to give you the desired output. For 2V out the I/V resistor becomes 27R and the LCR for the anti-sinc filter scale suitably. The E180F will have lower noise than the 6922 and NOS E180F are not difficult to get and do not quite carry the same silly pricetags as NOS 6922 style tubes.
Based on the Mullard Triode curves the operation will be 75V/12mA, gain is around 50 if CCS or Choke loaded. Output impedance will be around 2.5K.
So the Tango TC-160/15W Choke will be usable as Anode Load choke, otherwise Magnequest may have a suitable item or a CCS (IXYS 10M450 preferred) can be used. It is important to orient the chokes for lowest Hum.
Ciao T
This is wrong.
First, the Philips Datasheet is conservative. At AMR we did run extensive tests on that. The TDA1541 can handle more but not a LOT more. 50 Ohm I/V resistance is fine.
Secondly, if you want to stay within the 25mV compliance reliably for all possible output currents of the TDA1541 range you need 5.4 Ohm I/V conversion resistor (0.025V / 0.0046mA = 5.435Ohm).
My original "Adagio" DAC with TDA1541 used approximatly this value of resistance and a 1:15 Step Up Transformer. It also used an SRPP output stage tuned for lowest distortion, due to concerns about some of the ultrasonic "images" produced by the absence of oversampling folding back into the audio range.
While sounding good in the end we found the sound better without transformer and with a circuit that did not use agressive cancellation of even order harmonics. This then brings us back to the 2005 TDA1541 Analog stage.
For fun you could try using the E180F triode connected and with the I/V circuit scaled to give you the desired output. For 2V out the I/V resistor becomes 27R and the LCR for the anti-sinc filter scale suitably. The E180F will have lower noise than the 6922 and NOS E180F are not difficult to get and do not quite carry the same silly pricetags as NOS 6922 style tubes.
Based on the Mullard Triode curves the operation will be 75V/12mA, gain is around 50 if CCS or Choke loaded. Output impedance will be around 2.5K.
So the Tango TC-160/15W Choke will be usable as Anode Load choke, otherwise Magnequest may have a suitable item or a CCS (IXYS 10M450 preferred) can be used. It is important to orient the chokes for lowest Hum.
Ciao T
Hi,
Given the many configuration options and the options to use current or voltage output there are a lot of ways of doing this.
There is an unpublished article (it was submitted to enjoythemusic.com DIY Mag) that updates my old "Valve analog stages for Digital Audio" article from '99, it has several options that could be adopted...
Ciao T
Given the many configuration options and the options to use current or voltage output there are a lot of ways of doing this.
There is an unpublished article (it was submitted to enjoythemusic.com DIY Mag) that updates my old "Valve analog stages for Digital Audio" article from '99, it has several options that could be adopted...
Ciao T
Might not be a bad idea to post that as a blog if it is going to remain otherwise unpublished?
Hi,
It's kinda long and many need some rework, not very suited to blog. Will see how I do on time.
Ciao T
Might not be a bad idea to post that as a blog if it is going to remain otherwise unpublished?
It's kinda long and many need some rework, not very suited to blog. Will see how I do on time.
Ciao T
Hi,
I have nothing specifically worked out.
Given the outputs can be differential or parralel etc. it is hard to make "one schematic".
In principle just use suitable I/V resistors to Vref of the DAC and a normal tube stage. You will have of Vref worth of offet on the grid, so you need to use suitable cathode biasing.
There is really nothing special to the ESS DAC that does not apply to other DAC Chips.
The Manual for the Universal Tube Stage is on-line, it contains many configurations and also a principle schematic which should be quite clear on the principles. The exact values for parts etc... Well, it's easy enough.
Ciao T
I have nothing specifically worked out.
Given the outputs can be differential or parralel etc. it is hard to make "one schematic".
In principle just use suitable I/V resistors to Vref of the DAC and a normal tube stage. You will have of Vref worth of offet on the grid, so you need to use suitable cathode biasing.
There is really nothing special to the ESS DAC that does not apply to other DAC Chips.
The Manual for the Universal Tube Stage is on-line, it contains many configurations and also a principle schematic which should be quite clear on the principles. The exact values for parts etc... Well, it's easy enough.
Ciao T
@chanmix51
As is customary when all his technical objections are met ThorstenL attempts to move the focus of an argument to a subjectivist position with no verifiable foundation.
Take heart, however, from the following quote from Douglas Self:
'nods to Subjectivist convention are unlikely to damage real performance; this is however not the case with some of the more damaging hypotheses, such as the claim that negative feedback is inherently harmful'
Sorry all! my fault. Please take my posts with bags of salts (I've written many times that I am beginner).Thorsten, if it is not too much work, can you or someone else share what will be the new values of LCR filter in this case? Better how do you calculate them? Does this inverse sin(x)/x filter is like parallel RLC filter?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.