[mtlin12]

Introducing my new Cascode JFET+Diamond Buffer Pre-& Head-amp, Ultima DB-01.

Using DIYZONE's shunt regulator +/-24V for DB-01 with Relay PCB

stereo Ultima DB-01 Pre-& Headphone-Amp Module

Circuit Topology: All discrete JFET,BJT except LM334Z for C.S with no zener diode and aluminum Capacitors on the module.

The input stage is Linear Systems LS844 low noise Dual JFET.
This module was modified from my 150W MOSFET drive module.

http://myweb.hinet.net/home3/mtlin12/audio/articles.htm

 

[guzzler]

wow! that is a seriously professional looking job! and a very nice looking setup overall; good stuff!

g

 

[mtlin12]

Thank you, guzzler.

Actually Ultima DB-01 not only LS844 Dual JFET, but also use other
7*JFET, 2*CRD, 13*BJT, many 0.1% Philips resistors and Wima caps,
it cost a lot, but it is worth for DB-01's high speed, low noise and
high current drive capability (>200mA). It is the same
as my Ultima DT-01 MOSFET 150W Amp, all designed for SACD.

I've test them with my Marantz SACD DV8300, my headphone AKG
K-501 and my speaker B&W P6, it's better than my new tube Pre-Amp
and my old Perreaux 3350 300W MOSFET Power Amp.

DB-01 at 10K Hz Square wave input vs ouput

DB-01 at 100K Hz Square wave input vs ouput

DB-01 at 50K Hz 100 Ohm loading output 11.3Vrms

DB-01 at 50K Hz 100 Ohm loading nearly no phase shift

Very low noise about 10uV rms when input shorted .
DB-01 original bread board with DZ STAX-like shunt regulator.

 

[JohnFerrier]

Nice 1st and 2nd posts. What output transistors did you use? What VCC?


JF

 

[mtlin12]

I use normal MJE340/350 as output stage, it maybe not the best choice , but just the same as my second cascode stage,
when in DT-01 power amp mode I drive them at +/-68V, of course MJE340/350 can sustain much higher voltage.

When use in the DB-01 headphone-amp, the DZ STAX-like Shunt
regulator's voltages are +/-30V , and their currents are set at 280mA for both channels.

The other DZ's shunt regulator use OPamp instead the STAX-like
differential BJT, it is set at only +/-24V.

All pictures above are tested with STAX-like +/-30V shunt regulator. Because AKG K-1000 needs more power!

 

[ppl]

A Real nice design, there is nothing wrong with using an Op amp circuit topology as the basis for a discrete component design. I know of at least one well respected Amplifier designer that learned Amplifier design from Playing with discrete versions of Circuits found in Op Amp data Sheet simplified Schematics.

You might give some consideration to the Following Changes I would do to your circuit. (1) Add resistors to the base of Output and driver transistors to Kill the "Q" formed by Wire inductance and the transistors input capacitance. (2) using the lower current source to balance the Differing gain of the NPN and PNP Voltage gain stages. is a great idea and i have done this allot myself, However you could also leave that fixed and adjust the current source between the Bases of the transistors as an alternative.

Cascoding a Jfet to a BJT is or shall i say can be made into an Ideal Cascode. I would be interested in why you used a Jfet + BJT Cascode rather than an BJT+BJT or JFET+JFET Cascode?

Is that phase lead cap across the feedback there by default or did you have instability problems without it? if unstable with out the cap the circuit should be made stable without it and then if you still like use the cap to trim the Sq wave Response. IMHO phase lead compensation should not be relied upon to stop oscillation as this then is just a band-Aid used to fix an inherently unstable circuit. I do not like the use phase compensation cap's as they seriously affect the sound through out the audio range even though the effect of the capacitor is well up in to the MHz. range. Since it is your design the choice is yours as to the use of this cap or not just as long as the circuit is stable without it.

Good luck and happy listening to your Amp and please do share any further Designs you come up with on this Site.

 

[JohnFerrier]

Thanks for the pictures and information.

Is this a finished project, or do you have other ideas that you want to try?


JF

 

[mtlin12]

Thank you, ppl. Your suggestions are very good.

Let me explain why I use JFET+BJT, it simply wants to have some different designs with Erno Boberly's JFET+MOS , the BJT actually
should be a darlington pair, I use 2N5551/5401 with MJE340/350,
to get higher hfe.

2. I do not like cap in signal pass also, it is not because of unstable, I use 22pF to filter 200KhZ above, I think maybe can lower some noise, I use 2 other 22pF ceramic caps , one is for phase lead.

3.The CRD is choosed about 2mA, and I use High hfe BJT 2SC2240/2SA970 in the front that will lower the current from JFET.

4.The first stage current source, I even add a zero temp-compensate diode between the JFET and LM334, it is also set at 2mA.

5.Between the two Voltage source , I also design a stand-by circuit , that can stop the current source's working, but it must
be co-work with my relay board, it is not very useful at the Pre-amp, but very useful at the power amp. It also can be controlled by microprocessor.

 

[mtlin12]

Quote:

Is this a finished project, or do you have other ideas that you want to try?

I have finished two projects on Ultima-1 series, DT-01 and DB-01.

Now in Taiwan, they have been licenced to DIYZONE.

http://www.diyzone.net

I'm still plaining other products using this same structure module.
Like this:

using SANKEN BJT 2SC2922/2SA1216 Pc200W to have about 70W.

It is very easy to changed from 70W Class AB to 25W Class A design.


But first I just want to ensure the DT-01 and DB-01 to be
excellent products for DIYer.

 

[mtlin12]

It is time I should go to bed, see you tomorrow.

 

[erix]

Nice work!

You stole my fire with the stacked PCB's though! I have a couple of amps in the works using that idea.

Keep it up!

ok,
erix

 

[Superkiwi]

Bravo!

 

[steel]

That's a great job....

 

[jeffreyj]

Quote:

Originally posted by ppl ... You might give some consideration to the Following Changes I would do to your circuit. (1) Add resistors to the base of Output and driver transistors to Kill the "Q" formed by Wire inductance and the transistors input capacitance.

Mtlin, this is an excellent suggestion; alternatively, you can try slipping a ferrite bead over the Base leg; sometimes this works just as well and has the added benefit of counteracting some of the inevitable phase shift.


Quote:

...IMHO phase lead compensation should not be relied upon to stop oscillation as this then is just a band-Aid used to fix an inherently unstable circuit.

Ppl and I don't quite agree on the use of pole compensation (aka - phase lead, from when your only choice of amplifying device was a vacuum tube), but the point he makes above is a good one. If your circuit is oscillating, then it needs some help. Despite my evil love of pole compensation -

- I would ditch the cap in this case because the output buffer topology looks to be extremely well behaved. If there is any hint of misbehavior (as is, there is just the barest hint of underdamping in the first square wave response photo), such as leading edge ringing, then I'd try swamping the Miller capacitance of the voltage gain stage with 47-100pF. This is usually much less offensive to the most sensitive ears among us (which I *don't* claim to have, mind you...)

 

[mtlin12]

I watched closely at my Metrix OX 800 scope, and I found there
exist about 1.5MHZ oscillation in the first few minutes when powered on.
I doubt it was caused by the shunt regulator, so I
use the pole compensation cap to minimized this oscillation, I
think if there is no harm for the power Amp, user can take this 22pF off.

In my "paintbrush" layout PCB, there are many options for DIYer
to change the circuit, including the resistor to the base of the BJT.
Also different pin functions like 2SK214/2SJ77 and IRF610/9610
MOSFET can be handled, I want users to decide which component
will be the best.

In original design, DT-01 used MOSFET to be the source follower.
But in DB-01, I found that Diamond Buffer structure's speed is much higher
than the MOSFET, because of the Ciss, Crss of the MOS is much higher arround
100pF, so I decided to use Diamond output Buffer, very similar with the META42?!

Sorry, I must go to the church right now, see you.