I dunno how you'd get a DHT quiet enough (for some).
Parallel 5687 and Rp ~1k.. is good enough to forego the CF (for some, me included).
Why not use the -2mA DC offset into I/V resistor 27-39 ohm and direct couple (something like) E280F (cathode at 0V) into 2:1 for 2vRMS at 600 ohms = done.
Any perceived 'out of phase' issue can be easily corrected at the speaker (flip it).
Parallel 5687 and Rp ~1k.. is good enough to forego the CF (for some, me included).
Why not use the -2mA DC offset into I/V resistor 27-39 ohm and direct couple (something like) E280F (cathode at 0V) into 2:1 for 2vRMS at 600 ohms = done.
Any perceived 'out of phase' issue can be easily corrected at the speaker (flip it).
Last edited:
Sorry?
The many DHT preamps have made with success. Even for MM preamplifier people using DHTs. And there are few ways to handle the heating. I dont see any issue about using DH tube for this possition?
.
Cathode follower will introduce more complicated circuit. Yes it eill output significantly lower impedance.
Any way the using second section for CF is not the best option. Rather to put some different tube for that job.
But I think that 1020 ohm is low enough to drive most of the input stages.
I can cjheck how it works with 10K input R of nezt stage, because it is a value for some small % of DJT devices...
.
This is milion times already done approach which i dont see the point to perpetuate to infinity?
And there is common mistake in it:
First overamplify the signal to large value, with sound signature of no need fo amplification factor, then step down the votage to lower signal... Only benefit could be the lower impedance.
BUT everything else is wrong - for transformer usage best way is 1:1 because of prcticaly no squared transfer of "everything"...
sqr(1)=1
And again that way is good but ONLY if the Riv is chosen for min. THD, tube is chosen respect to Vp-p @ Riv and some normal desired output p-p value. And 1:1 isolation transofrmer used. Without reactive transfer back and forth...
.
Sorry I dont want to be harsh please, but i think that is not good at all. Earlier some "rulle" was standard - phase in =phase out from the device...
Now people switching the terminals of speakers...
.
And why it is so forgiving about the phase inversion? Almost as it actually better to have reversed phase than normal - in hase?
Yes, I get it. At least in the sim and with the 5687 Models I have in TINA, the HD is basically that of the current conveyor.
So current conveyor and tube as cascode will have very low HD.
Lack of availability of such tubes?
Actually, the I/U conversion should be inverting for TDA1541, It is non-inverting circuits that need inversion somewhere.
If you care about that sort of thing. Now some studio microphones have inverting gain stages after the capsule followed by high ratio stepdown matching transformers, others have followers after the capsule and low ratio stepdown or unity gain transformers.
Which microphone has the correct polarity?
Thor
One of the I/V stage modular pcb's will be with D3A. I don't mind making any design you guys deem solid for this dac. Grounded grid can be made to accept variety of tubes (cheaper ones), with little change in sound, so one like that as well. Solid state one will be virtual zero psrr thorstens design. Also one i designed for pcm63. Thats the beauty of it. You get tired of one, want to change it up a bit, thats the beauty of modular design. Also it saves you much of the trouble if/when tubes you like go completely obsolete/prices go trough the roof. While still looking good 😁
Yes.
But there is one more topology employing the tube.
Sort of "memory" circuit but instead of JFET there is a triode.
Unfortunatley this could be done only in "positive" branch of supply.
I will send some concept sch.
Uh, with all the respect due to you, here I flee (and what I have seen critising tube circuitries in the ad1862 thread encourages me HERE to do so twice) .
Sorry for the noise. I encourage to follow T circuitries for both tubes and BJT, and Zoran for BJT (and tubes too).
(nothing personal). I even see some thinking there is a riussian technology about cheap East copies of ECC88 (w/o knowing that) or even worse with E188CC when said russian equivalent had much more microphonic because figthing technic are cheap in their tube construction! Some even tried to maake believe us than used Philips tools exported in East Europe sounded better ! Wake up guys !
If it's to go for tube to make wosre than BJT w/o the tubes subjectives goodies, better tosave monney (imho)
I don't mind too 😉
Last edited:
There are now complete shematics to test one can test verroboard or more expensive with 6 layers pcbs.
It will be good to test them first before talking about the color of the car, imho, YMMV. Rest is just car options, noise non ruled one, and finally noise here !
I modstly think, all bad tubes from China, Russia, w-/o shematic that can compensate them because dispersions in all the direction die to the bad fab, is just.... not deserving what tubes (goods') can bring.
Thorsten with all his experience said BJT current conveyor in his ciruitry can do best. Zorzn who is not the last here in audio electronic and know both, focus on BJT on most of his answers.
If we can first legit one shematic or two before talking of the rest, I believe the noise ratio of the thread could profit of this !
It is not about gurutry, it is just about knowledhe and experience some clearly have more than others !
mode off !
It will be good to test them first before talking about the color of the car, imho, YMMV. Rest is just car options, noise non ruled one, and finally noise here !
I modstly think, all bad tubes from China, Russia, w-/o shematic that can compensate them because dispersions in all the direction die to the bad fab, is just.... not deserving what tubes (goods') can bring.
Thorsten with all his experience said BJT current conveyor in his ciruitry can do best. Zorzn who is not the last here in audio electronic and know both, focus on BJT on most of his answers.
If we can first legit one shematic or two before talking of the rest, I believe the noise ratio of the thread could profit of this !
It is not about gurutry, it is just about knowledhe and experience some clearly have more than others !
mode off !
Last edited:
It's the beauty of it, one can use what one prefers. Or even better since it's the name of the game, try different solutions and pick the best. All with ease. D3A is very linear, high gain and very low noise tube, all boxes checked for a good I/V stage. Many will be made, all good, all different flavours 🙂
The best version of grounded grid is with ECC88/PCC88 parallel sections and 2 x 2SK2394CP6.
Not so friendly to lower signal values (lower Riv) thd rise with lower Riv at anode.
(I tested in pspice with very accurate models. Performed better than 8020 and other xpensive tubes...
Also many JFETs are checked contemporary and old, and best results are with 2SK2394CP6. CP7 version i think it is not available any more?)
.
But these latest versions has better PSRR and THD, no chage of THD versus Riv values
All that with a few elements more.
.
Off-course not hard to try diferent options of IVs with open PCB modular design... 🙂
Last edited:
Yes, exacly the point 😁 It's diy. Not a commercial product. So many options of flavours to each ear/brain.
Not as much as other high gain tubes. This one is as silent as high mu tube can be, and not really an oscillation beast. One downside is, it's becoming very rare, and very expensive. And once your stock is gone, there really isn't any replacement for it...
Yup those tubes used to cost not so much, then pople purchased names more than circuitries... fame factor. Or people were not loving some circuitries because avoiding ferites beads when needed thinking it was not then said the tube was not so good sounding....
At the end if the two channels have the same gain, why not indeed ! After all, here some current conveyors have been proposed since few pages. But it is just peope are buying Alixpress crap kit or tubes thinking it is the same because it is labeled... tube !
Again, take Taylor swift with short legs full of furs and you can be desapointed, not having what you were promised! Possible but risky, according the tube, circuitry, etc ! Anyway if people have choice this is okay, sure ! 🙂
At the end if the two channels have the same gain, why not indeed ! After all, here some current conveyors have been proposed since few pages. But it is just peope are buying Alixpress crap kit or tubes thinking it is the same because it is labeled... tube !
Again, take Taylor swift with short legs full of furs and you can be desapointed, not having what you were promised! Possible but risky, according the tube, circuitry, etc ! Anyway if people have choice this is okay, sure ! 🙂
Last edited:
Seems best choice will depend on overall gain structure and actual requirements for 'low noise'.
I liked the simplicity of the parallel 5687, lack of CF and reasonably low Rp.
For me, that it has been done a million times is a good thing.
I liked it CCS loaded, source follow output through a coupling capacitor, tried the transformer and it was different and better in most ways, a lot simpler, less to go wrong and less to troubleshoot. Just easier. 1:1 would have been nice but I need lower than 2k output impedance and could afford to give up half the voltage gain. For me, 2vRMS at 600 ohm is about right. Obviously, these things are situation dependant, and with personal preference etc.
Ah, no problem.
Personally, I don't have an issue with this 'correct' phase thing. Perhaps my ears cannot notice, or are just not bothered by it.
Yes, although somewhat mitigated in the sense that D3a, E180F, E280F and E810F are all pin compatible and similar enough.
Prices of E180F/6688 are comparatively low and availability is good.
Yes
anyway the increased dynamic capacitancy is not the issue because Riv alredy has parallel C for HF filter. So the Cdyn can be subtracted from "CiV".
.
And more important:
that 1/2 Ri for parallel mode is reducing the primary inductance of optionally 1:1 output transformer.
To feasibile value, for good LF response.
I can check but I think that the 30Hy will be good... 🙂
.
The key for "new" approach is to have not so high amplification stage. but middle to lower -A total.
To avoid sound signature and have better PSRR and better more neutral harmonics order.
So the diskrete IV is a sort of current buffer in this case.
With "isolated" Riv from DAC output. DAC is not a current driver for low impedance sources like small values of Riv-s
.
More: The Riv in BJt IV circuit could be exchaged with transformer.
So as the IV circuit using the DAC Power Supply, it become part of DA segment.
With tranformer in the Riv palce, we van isolate the AD segment from pure A segment.
So the tube with own power supply in that case is pure analog stage.
.
Off-course in that case we loose direct negarive grid bias and bring back 2 elements Rk and Ck in the cathode... 🙂
But coupling C is not needed.
.
Thw same happening in the Digital side, and that is why all the inputs should be isolated with digital isolators.
As well as external dem source...
This way DAC as AD segment is isolated from pure Digital segment. Digital segment has own PS.
.
For this concept we need 3 separate transformer cores. To avoid capacitive coupling and othet interferences that become fact with 1 or 2 cores power XFRMs...
.
And digital isolation can be used as seletor of digital inputs. IL710 has enable input, so with enable/disable digital isolators we can get SPDIF / USB digital sources selection.
I did it many times and it wotking very good.
Yes
anyway the increased dynamic capacitancy is not the issue because Riv alredy has parallel C for HF filter. So the Cdyn can be subtracted from "CiV".
Can you explain in different words please?
anyway the increased dynamic capacitancy is not the issue because Riv alredy has parallel C for HF filter. So the Cdyn can be subtracted from "CiV".
Can you explain in different words please?
I think 4 Layer is doable, TDA1541 is pretty primitive.
D3a is a great low level stage tube in both Triode or Pentode and a great single stage driver tube for non-NFB Amplifiers.
My take would pentode (with CCS based servo screengrid PSU to prevent dynamic current "escape") with a transformer load, perhaps 2:1 (2.4k:600R) for 2V out at 600R source.
They are now rare and are pushing beyond 100 USD per tube. I can get 6AK5 NOS for under 10USD. GE 5670 for under 20 USD.
They are not as linear. Not as high current or gain. But I don't spend 200 Bux on tubes for 2 Channels.
Thor
Beware. Do not use the RL timeconstant for -3dB. Take the lowest frequency of interest and the net impedance in the primary circuit, keep current modulation at 20Hz (or 10Hz if you wish) to no more than 1.41414 times the signal current at 1kHz,
Given 1.5k appx source impedance, we are looking at a minimum 4:1 stepdown if we want a 600R load.
We will end up with a parallel feed (capacitor coupled) transformer for enough primary inductance.
Doable.
A TDA1541 Stage from my drawing board used a very large step-up (1:40), differential 5687 (or similar medium mu Dual) and Peerless 217 Outputs (Cinemag replica) with cathode feedback. I eventually decided that this was not "the way", for me at least. Loads of bragging rights.
Even better, we could use dual 101D DHT's with a few small changes and then a differential TDA1541.
It would sure make a good "Hai Enten" DC to sell for 20,000 USD a piece to the Hai Enten Pee-pull, as long as it has an asynchronous LC Oscillator with a 102D DHT to provide DEM clocking and Mundorf Silver Oil capacitors for the separate power supply..
Hence my relatively simple BJT based circuits, as "Hai Enten Antidote". Ask D Junior Trump about Batteries and Sharks. Hai Shark in German, Ente Duck - so Hai Ente liguistic joke for High End)
The discrete circuits I have showed are not I/U converters.
They are merely current conveyors, meaning ideally 0 Ohm input impedance, infinite output impedance and zero "current escape", so all current at Iin exits at Iout.
The I/U conversion is done by a resistor.
Indeed. But then the I/U conversion is still done by resistor.
If the digital segment has it's own supply and inputs are isolated, digital isolators are superfluous and add unneeded complexity, for complexities sake, with at best only no performance disbenefit.
Again, overcomplication, where screens suffice.
Thor
re: D3a at 2k4 Rp is correct - the charts suggest this for 10-15mA and close to zero bias, confirmed with uTracer (real world) across many samples.
E810F has half the Rp and appears to be the outlier amongst that grouping, all else same and Gm doesnt go through the roof until you crank the current.
I can find 200pcs E180F at USD 8 ea, all day (today or tomorrow). Min qty = 1. Which suggests you can probably find them at that rate also.
E810F has half the Rp and appears to be the outlier amongst that grouping, all else same and Gm doesnt go through the roof until you crank the current.
I can find 200pcs E180F at USD 8 ea, all day (today or tomorrow). Min qty = 1. Which suggests you can probably find them at that rate also.
Yeah, at 100...150 Bux, a bit much.
Yes, D3a and E810F are among my favourite tubes. And they cost same these days.
E810F has more transconductance, D3a more Mu.
Can we current manufacture to original Siemens standards for say 30 Bux each?
Yeah, but there is no real difference to the 6AK5 in most applications, which are cheaper.
Thor