F5X -- the EUVL Approach

[lhquam]

I am interested in a group buy for Toshiba MOSFETs. We should start a GB thread for this.

Originally Posted by EUVL

I have been asked by PM about sourcing MOSFETs.

You probably all know that Toshiba is stopping all audio devices, MOSFETs and JFETs alike.
The only open source that is known to have sufficient quantities for NNPP match is, as mentioned before, ZhouFang.
It has also been discussed before that his matching is not exactly at F5X operating conditions.

So if you do not wish to purchase from Zhou, the only solution right now is to GB and match yourselves.

We ourselves are in the process of acquiring sufficient devices to support future batches of F5X projects.
But it won't be 6 months or so before we are ready, and then only in the context of supporting the project,
and not as separate devices sets for general sales.


Patrick

[lhquam]

I am working on a variant of Patrick's F5X board that is shrunk to a 2.5" x 3.8" board to be mounted on a 5" x 7" heatsink . Two such heatsinks are attached fin-to-fin to form a heat tunnel, cooled by a very quiet 120 mm fan. Here are images of the layout:

The first image shows the entire heatsink, with the airflow from top-to-bottom (yeah, I know, this is upside-down from normal conventions). The MOSFET placement was carefully chosen for heat distribution. The design is for 100-120 watts dissipation per heatsink.

The second image is the PCB. The resemblance to EUVL's layout should be apparent. Due to space restrictions, the 220 ohm feedback resistors are replaced with Caddock MP820's attached to the heatsink.

There are 2 major jumpers (not shown) for the X connections. Without any other jumpers, the board provides an F5X. Two jumpers provide a floating balanced togolopy, (the X's are connected). Two more jumpers activate Nelson's P3 adjustment. And finally, two more jumpers ground everything for initial adjustment or for use as 2 independent unbalanced F5 amplifiers. The jumpers are intended to be made using .1 inch shunts, but soldering is possible.

Balanced input connections are made using a 4 pin header. At the corners are 2 pin headers for connections to the source resistors for initial adjustment.

[pchw]

There is a thread for Toshiba 2SK1530/2SJ201 GB discussion now.

[EUVL]

There were some discussions about distortion cancellation of the F5X at the build thread.

The original discussion in 2008 (!!!) you can find here.

F5 power amplifier

Fitzfish and I also stated on a couple of occasions that the measured distortion spectrum does not differ between ground X, floating X, and the H connected (see NP's article on F5 Turbo, the balanced schematics towards the end). However, the floating X or H will allow the circuit to operate with single ended inputs (-Vin connected to Gnd), which is not the case for the grounded X.


Patrick

[alazira]

Hi,
If I have a choice between two match quads of 2sk1530/2sj201
0.265 / 0.265 / 0.265 / 0.265 or
0.350 / 0.350 / 0.350 / 0.350
which, if any, would be better for this project?

[pchw]

Assume the numbers are the Vgs different between N and P, I suppose closer is better, i.e. 0.265 / 0.265 / 0.265 / 0.265. Whether there will be audible different between the 2 sets is another story.

[EUVL]

You need to show me actual Vgs (also at what current and what Vds) of all 4 MOSFETs in a set, before I can tell you which is better.


Patrick

[alazira]

Yes, they are Vgs, matched at 24V and 1.3A (for standard F5). If possible I will test at 2A but was just wondering if all things being equal, would the lower or higher Vgs be advantageous for the F5X.

[EUVL]

Are they 2SK1530Y / 2SJ201Y ?
If so I expect the Vgs to be around 2.xxV, and not 0.2xxV.


Patrick

[EUVL]

Now I know where these numbers came from.
If I were you I would choose the higher numbers.


Patrick