designing PCB for Balanced Bridge Gilmore Dynamic (preview)
Sep 25, 2003 at 7:35 PM Thread Starter Post #1 of 9

Glassman

Headphoneus Supremus
Joined
Mar 22, 2003
Posts
1,830
Likes
11
I'm designing PCB for balanced version of Gilmore including Elliot's capacitance multiplier PSU on the standard dual-sided Eurocard board (160x100mm).. just to let you know
tongue.gif


here are the holes placed as they'll be on the real board and comparison with unbalanced pcb from mcnowinski based on sijosae's work.. the opamp servo will be dual-opamp so that one chip will serve both channels..

BBGD-preview.gif


I appreciate any comments or tips on improving the amp.. do you think that such capacitance multiplier will be better than lets say LM317/337
confused.gif
 
Sep 27, 2003 at 11:30 PM Post #2 of 9
Quote:

Originally posted by Glassman
I appreciate any comments or tips... do you think that such capacitance multiplier will be better than lets say LM317/337
confused.gif


I recommend revising your design to use the Linear Technology LM1963 low noise regulators. These have a noise rating of < 40uV rms. The lower current version have < 20 uV rms. Checkout the datasheet for the 1.5A version here: http://www.linear.com/pdf/1963af.pdf

Basically, LT and Texas Instruments have incorporated the Jung Super-regulator idea into these parts. They have been available for a few years now. I don't know why there isn't a bit more discussion about them and I certainly don't know why people continue to use LM317s, if they they are trying to build the best.

I'm going back and forth over a bridge design. I'm not sure the benefits out weigh the extra complexity, more noise (~3x), and possibly more distortion (though I guess even harmonics tend to cancel). I do like a symetrical design. Though I don't think I'll build Gilmore's, I like it a lot.

Good luck with your design!

JF
 
Sep 28, 2003 at 12:00 PM Post #3 of 9
thanks for suggestion, I'll try to get some samples of that regulator.. it looks pretty good
wink.gif



back to Gilmore: one of my friends suggested me a possible improvement on original Gilmore circuit - adding a current mirrors instead of those 5K resistors.. do anyone know how to count this?

and also I found a nice circuit which looks almost exactly like Gilmore
smily_headphones1.gif
here it is..

almost-gilmore.gif
 
Sep 28, 2003 at 12:44 PM Post #4 of 9
quote
back to Gilmore: one of my friends suggested me a possible improvement on original Gilmore circuit - adding a current mirrors instead of those 5K resistors.. do anyone know how to count this?

You can certainly cascode the input and the second stage, but
you loose voltage swing when you do so. At +/- 15 volt
power supply levels you can loose half your output.

I've never seen that particular circuit, i would like to know
the actual date on it. In fact most complementary output
power amps in the world look like this. Including mine.
Even as far back as the lexcon's (the brittish ones), early GAS...
The circuit has been around for ages and ages. There are
only so many ways to skin a cat.

What i did was make the output stage pure class A thus
simplifying the bias drive which never works absolutely
perfect no matter what you do. Removed all input caps,
made the circuit with a bandwidth wide enough to make
any compensation unnecessary, and changed the way
the servo worked. Evidently many more people than i
thought have found the led/transistor current source thing
works well, i found it in an obscure write up that was
20 years old.

In the design you posted i'm not sure why they think they
can get away with only pulling input currents to -15.
The increase bias current of the mat02 cannot be guaranteed
to be more than that of the mat03.

I'm working on a new version of my circuit now.
Cascode in both the input and second stage like the
uberamp. Pure class A output, multiple transistor thing.
+/- 30 volt power supply rails. 10 watts of class A bias.
Guaranteed to put 20 volts rms into sennheisers. (>1 watt)
Same thing with grado's. (>10 watts)
Should shut up the one person who claims my amplifiers
are not loud enough. Actually it should make him so deaf
that he won't be saying much to anyone. His problem, not
mine.
 
Sep 28, 2003 at 1:13 PM Post #5 of 9
you made me cry for it! what's the ETA? are you goin' to give us the new circuit as a christmas gift
biggrin.gif


will you stay with 2SJ109/2SK389 in your new design?

btw. me and my friend, we are going to build a small SMD-based version of your amp with DIP-styled legs as a discreete opamp and use it in friend's Kenwood CD player - we want to make a challenge - Gilmore vs. OPA627B
smily_headphones1.gif
 
Sep 28, 2003 at 4:25 PM Post #6 of 9
quotes
you made me cry for it! what's the ETA? are you goin' to give us the new circuit as a christmas gift

will you stay with 2SJ109/2SK389 in your new design?


I'm sure it will be a christmas gift. It is just the year that
is at issue. I doubt i will have it ready for release for 2003.
Way too much software to write at work, and as i get older
software messes with my brain too much. I hate sql, really
i do.

I'm desperately looking for ultra low noise complementary
devices for the front end. It is hard to beat the 2sj109/2sk389
combo. I am certainly willing to try any other ultra noise
matched pair's anyone is willing to suggest. Including
transistors.
 
Oct 17, 2003 at 9:06 AM Post #7 of 9
Quote:

Originally posted by kevin gilmore

will you stay with 2SJ109/2SK389 in your new design?

I'm desperately looking for ultra low noise complementary
devices for the front end. It is hard to beat the 2sj109/2sk389
combo. I am certainly willing to try any other ultra noise
matched pair's anyone is willing to suggest. Including
transistors.


-Linear Integrated Systems-
Ultra Low Noise Low Drift Dual N-Channel JFET
http://www.linearsystems.com/datasheets/LS843-5.pdf

NF: 0.5db Max.
Ciss: 8 pF Max.

Unfortunately, no matching P-Channel. Darn.

They also mfg LS318 (MAT02 equiv.) and LS358 (MAT03 equiv.).
http://www.linearsystems.com/datasheets/LS318.pdf
http://www.linearsystems.com/datasheets/LS358.pdf


JF
 
Oct 17, 2003 at 12:30 PM Post #8 of 9
Hi again John. The Linear integrated system transistors you recommend are real great I have used them a lot in my Designs for years. They were a spin-off of Advanced Micro Devices. They have some nice complementary BJT matched devices I like the LS-312 and LS-352 PNP/NPN devices as complimentary Differential inputs. The MAT-01 and MAT-02 are also real nice and I have had great luck with them. The Mat-XX devices have a diode between the base and Emitter that will protect them from reverse Bias, where as the LS-XXX devices do not. The MAT transistors sound real good however IMHO the sound is improved with the LS types I mentioned. The LS-840 Dual J-FET is one heck of a device. This device makes the finest sounding Cascode current sources for Biasing op amps I have used. Sure wish there was a P Channel Equivalent. I have also used the LS-843 as a Differential Input of the conventional non complimentary Topology typically used in most J-FET input Op Amps.

Kevin I use the 2N5457 N channel and 2N5462 as the P channel device in my own Complimentary J-FET Differential amp Input stage. I Cascode these Complementary J-FET’s to a pair of LS-312 and LS-352 Dual Complimentary BJT’s. I do not know if these are any better than the 2SK389 and 2SJ109 specified in your Amp or not.
http://www.onsemi.com/pub/Collateral/2N5457-D.PDF
http://www.onsemi.com/pub/Collateral/2N5460-D.PDF

The LS-312 and LS-352 are not listed in the datasheet Directory John linked to However on Linear Integrated Systems home page you can Find them by using that sites search. These I use because they are 60 Volt devices and thus are suitable not only for headphone and Line level Amplifiers but can Be cascaded and used in High Power Loudspeaker Amplifiers of 50-200 Watts
http://www.linearsystems.com


smily_headphones1.gif
 

Users who are viewing this thread

Back
Top