Building the ultimate NOS DAC using TDA1541A - Page 349 - diyAudio
Go Back   Home > Forums > Source & Line > Digital Line Level

Digital Line Level DACs, Digital Crossovers, Equalizers, etc.

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 18th October 2010, 11:42 PM   #3481
diyAudio Member
 
Join Date: May 2006
Location: Holland
Hi EUVL,

Quote:
As shown, the output impedance in balanced mode is 2x (1/Yfs) of 2SK1530, which will be around 1 ohm at 2A bias. For me rather too high. I suggest you look into devices like 2SK3497, maybe 2x parallel at 1A bias each, per circlotron side.
with 2SK1530 I measured 0.6 Ohms output impedance with 150mA bias current and no global feedback.

The straight-forward 4-FET design is possible because of the relatively low 2SK1530 MOSFET input capacitance of only 700pF.

The 2SK3497 has specified 2500pF input capacitance (and 5000pF for two in parallel). This would require an extra balanced driver stage, adding at least two extra semiconductors that would further reduce resolution.

Possible alternative for the power MOSFETs could be Silicon Carbide power JFETS from SemiSouth with specified rdsON of 0.15 Ohms and around 600pF input capacitance.
  Reply With Quote
Old 19th October 2010, 01:27 AM   #3482
EUVL is offline EUVL  Europe
diyAudio Member
 
Join Date: Oct 2003
Maybe you have forgotten that in follower mode, only Cgd counts, which is 100pF fpr 2SK1530, and 30pF for 2SK3497.

SemiSouth has about 100pF Cgd, and similar transconductance at low bias to 2SK1530.
For me output impedance too high.

Rdson has no meaning in follower mode.


Patrick
  Reply With Quote
Old 23rd October 2010, 08:47 AM   #3483
diyAudio Member
 
Join Date: May 2006
Location: Holland
Hi EUVL


Quote:
Maybe you have forgotten that in follower mode, only Cgd counts, which is 100pF fpr 2SK1530, and 30pF for 2SK3497.
Yes, seems I forgot this. I will test the 2SK3497 if I can source them without having to buy 100pcs.

There were some problems with the Circlotron prototype, two 2SJ109 JFETs blew-up (gate-source breakdown of the JFET connected to the input). I suspect limited breakdown voltage of 25 ... 30V.

Afterwards I am glad this happened as I put 200V, high Yfs power MOSFETs in the input / driver stage to prevent this from happening again.

The input capacitance is slightly higher (120pF) compared to the 2SJ109 (95pF). Yfs is specified at 35 (typical), 2SJ109 is specified at 22.

However, the perceived sound quality was cleaner and "stress-free" compared to the 2SJ109.


Quote:
SemiSouth has about 100pF Cgd, and similar transconductance at low bias to 2SK1530.
For me output impedance too high.
Main reason for me to test power JFETs is the resulting perceived sound quality. Devices that provide similar output impedance can still sound very different.
  Reply With Quote
Old 23rd October 2010, 11:14 AM   #3484
EUVL is offline EUVL  Europe
diyAudio Member
 
Join Date: Oct 2003
You can start a GB for 2SK3497, and 100 pcs will be gone by the end of the day.
You can count me in for 20.

In my design, I cascode the 2SJ109 with J174s. Worth a try. You can also use ZVN3310 with a zener at the gate.


Patrick
  Reply With Quote
Old 30th October 2010, 11:34 PM   #3485
brubeck is offline brubeck  Netherlands
diyAudio Member
 
Join Date: Sep 2009
Location: Netherlands
Hi all,
Today i was at John's house hearing the first tones of his new designed Circlotron amp in experimental stage. Look at the attachment.

It's not finished yet. There are still some variables and design philosophies to try. What you see is one channel.

After some listening it is clear to me that this amp has a really great control ability. i only heard this with digital amp's. This aspect is truly amazing! and i think a novum in SS-amp world. While the signal pad is so short without any signal capacitor the resolution is also very, very good. The flavour of sound is tube like, but with lots more power. The high tones could be more brilliant as tubes. We discussed about that and it's possible influenced by the type of PS-capacitors. John uses Panasonic FM at the moment. I prefer Nichicon Gold Tunes. What's next is to try a tube driver instead of SS to get some more 'tube magic'

This amp design is very promising..
Attached Images
File Type: jpg IMAG0089.jpg (587.8 KB, 920 views)
  Reply With Quote
Old 1st November 2010, 08:42 AM   #3486
diyAudio Member
 
Join Date: May 2006
Location: Holland
Hi EUVL,

Quote:
You can start a GB for 2SK3497, and 100 pcs will be gone by the end of the day.
You can count me in for 20.
Digikey hasn't got 100pcs in stock, estimated delivery is in January 2011.

2SK3497 has higher Yfs compared to the 2SK1530, this should lower output impedance, but it also increases local negative feedback. The increased negative feedback could result in more sterile sound.

Quote:
In my design, I cascode the 2SJ109 with J174s. Worth a try. You can also use ZVN3310 with a zener at the gate.
This conflicts with the principle of fewest possible semiconductors in the signal path. I managed to reduce total amount of semiconductors in the signal path to only 4. This is the minimum amount of components required to construct a balanced bridge amplifier with a gain of 30.

For some reason, low wattage JFETS / MOSFETs in the input / driver stage seem to lead to reduced perceived sound quality. I am now using P-MOSFETs (TO-220 package) with rated 1.75 watts and Yfs of 35 that are designed for audio and video amplifier applications.
  Reply With Quote
Old 1st November 2010, 08:46 PM   #3487
diyAudio Member
 
Join Date: Apr 2007
Reducing the number of the semiconductors in the signal path (hopefully down to zero) is the right direction. I'm sure that at the end John will rediscover the vacuum tubes as he did with the 1 DAC chip vs 8/16 such.

John, please give a chance to the good old 6N6P (6Н6П) even in Class A - Troels already rediscovered them (TRAM w.) or try something like ECC81. The tubes have no thermal (resentful) memory...
  Reply With Quote
Old 2nd November 2010, 10:01 AM   #3488
diyAudio Member
 
Join Date: May 2006
Location: Holland
Hi brubeck,

Quote:
We discussed about that and it's possible influenced by the type of PS-capacitors. John uses Panasonic FM at the moment. I prefer Nichicon Gold Tunes. What's next is to try a tube driver instead of SS to get some more 'tube magic'
When you visited me, Circlotron design wasn't completed, in fact I removed some components during experimenting (capacitance multiplier caps) and these weren't put back when you listened to the Circlotron. I also used some bulk metal foil resistors for experimenting that turned out to degrade performance.

But worst of all, the power supplies missed what now turned out to be the the most important circuit in the entire power amplifier.

Circuit performance depends heavily on power supply quality, regardless of how well a circuit is designed. Most important with Circlotron power supply is the use two truly floating power supplies with minimum stray capacitance between both power supplies.

I had to find out the hard way that plain semiconductor rectifiers are basically unusable for high performance designs as they dump switching interference on the connected circuit. This interference consist of pulses with a fundamental frequency of 50, 60, 100, or 120 Hz plus higher harmonics that cover the audio spectrum. The effect of this interference is a muddy "nervous" sound and a background that isn't black.

What happens is that peak charge currents occur, every time the rectifier suddenly starts conducting, and that back emf pulses occur (transformer) when the rectifier suddenly stops conducting and the transformer winding is essentially being disconnected while the voltage hasn't reached zero. Schottky diodes won't solve this specific problem.

The peak charge current is affected by smoothing cap ESR, thus caps with very low ESR will result in highest peak charge currents and highest switching interference levels. Theoretically, low ESR caps should lead to poorer sound quality as they produce higher interference levels when used in a conventional power supply circuit. Capacitors with higher ESR would perform better here as peak charge current would be lower, thus interference levels would be lower.

Unfortunately, circuits perform best when the power supply impedance is as low as possible. Here low ESR (Equivalent Series Resistance) helps.

In practice, the capacitor that offers best balance between impedance and switching interference would offer best sonic performance. Capacitor "rolling" can lead to acceptable performance in a specific application, but it cannot fix this fundamental flaw in power supply rectifier and smoothing circuits. Note that once the interference is generated it's virtually impossible to fully remove it.

Some time ago I published the stepped rectifier concept. It's a simple cheap circuit consisting of some diodes and resistors.

Here the charge current is both, increased and decreased step-wise. This is done by using diode threshold voltage in combination with specific series resistor value. The diode forms an electronic switch that puts the resistor in the current path (diode does not conduct), or bypasses it (diode conducts) as the primary voltage rises. Multiple diode-resistor switches can be placed in series to control the charge and discharge curve. This way the problem is tackled right at the source, by minimizing switching interference.

Now it's no problem to use very low ESR caps while keeping switching interference levels very low.

Yesterday I installed stepped rectifiers on all Circlotron power supplies (including driver stage power supply). I put switches on these stepped rectifiers to bypass them for comparison. With the stepped rectifiers disabled (short circuit) the perceived sound quality is already very high. When the switches are opened, the last bits of "stress", harshness and grain completely vanish, and the sound becomes very realistic. Compare it with listening to music in a room with all kinds of background noises vs a dead quiet room.


Quote:
What's next is to try a tube driver instead of SS to get some more 'tube magic'
First I plan to complete Circlotron MK1 PCB design, the stepped rectifier mod had such impact that I am quite happy with the results of this all-MOSFET design.

Last edited by ecdesigns; 2nd November 2010 at 10:05 AM.
  Reply With Quote
Old 2nd November 2010, 11:37 AM   #3489
diyAudio Member
 
Join Date: May 2006
Location: Holland
Hi PyroVeso,

Quote:
Reducing the number of the semiconductors in the signal path (hopefully down to zero) is the right direction.
Most designers use lots of semiconductors in their design to improve static performance. Static performance can be verified / measured, and the assumption is made that this leads to better sound quality. This is true to some degree, but when it comes to refinement, these static parameters are of little use.

In order to achieve best tracking, servo systems require lowest possible backlash. Example, a milling machine with worn out gears won't be able to mill accurately despite servo correction mechanism. It can still operate with reasonable accuracy, yet fails to maintain high accuracy. In other words, when backlash increases, smallest movements / positioning is affected first.

Every semiconductor has "electronic backlash" comparable with backlash in a gear system. This "electronic backlash" of a semiconductor determines the smallest voltage fluctuation it can track.

Using many semiconductors in a design increases the risk of increased "electronic backlash" and resulting poorer accuracy / tracking. This is why I attempt to minimize the amount of semiconductors in the signal path as this is the simplest way to ensure lower "electronic backlash".

Quote:
The tubes have no thermal (resentful) memory...
This is why I use "power" MOSFETs in the combined input / driver stage instead of low wattage JFETs like the 2SJ74 or 2SJ109 for example.
  Reply With Quote
Old 2nd November 2010, 11:50 AM   #3490
diyAudio Member
 
Join Date: Feb 2005
Location: Melbourne
Yes, Thanks John, for this "Stepped Rectifier Network" design - such a simple thing that works like a charm!

I used it in the headamp and added a single R-C "snubber" and very much quieter amp. Had same success with 15ETH06s and BYW29 diodes in the power amp even with just the 3 'steps'.
__________________
... jh
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off



New To Site? Need Help?

All times are GMT. The time now is 09:31 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2