LuDEF

Jump to Latest
#423
woofertester, after you've had a chance to measure some of the LD1014Ds, could you please give an estimate on how many would be needed to be purchased to get a reasonable matched pair...assuming one LD1014D per channel.

(Might need a closer match if lhquam comes up with a paralleled monster ;-)

Thanks for all your work!
 
#424
depending of circuit, one wants to use them in, matching can be even irrelevant

I believe in source follower stage (common drain), same as in LuDEF, transfer characteristic difference is pretty much trivial

and then, if there is some loop NFB (well, not case in LuDEF, existing one) , all differences outa window

now, common source stage, without NFB, different thing
 
#425
lhquam said:
Here is a comparison of my LU1014D measurements vs. the SPICE model from Zen Mod posted at post #178 LuDEF.
Other than the Vgs range being offset (to be expected) they look quite similar,
Click to expand...

If you have a way to perform remote sense on Vds, the curves get sharper and less resistive looking. I will post an A-B comparison when some connector parts arrive. Unless, of course, you already are doing remote sense voltage measure at the device terminals.
 
#426
GnuB said:
woofertester, after you've had a chance to measure some of the LD1014Ds, could you please give an estimate on how many would be needed to be purchased to get a reasonable matched pair...assuming one LD1014D per channel.

(Might need a closer match if lhquam comes up with a paralleled monster ;-)

Thanks for all your work!
Click to expand...


I agree with ZM. It depends on what you want to build. I match parts because I can not necessarily because it is required.
 
#427
woofertester said:
If you have a way to perform remote sense on Vds, the curves get sharper and less resistive looking. I will post an A-B comparison when some connector parts arrive. Unless, of course, you already are doing remote sense voltage measure at the device terminals.
Click to expand...

Below is the schematic for my "FrankenTracer" setup. For the LU1014D I use only one of the 24V transformer secondaries to get 24V RMS with 120V RMS into the primary. The Xout and Yout signals go to a Rigol DS1054Z Digital Sampling scope and the Vgs input to the SIT gate comes from a USB controlled Rigol DG1022A signal generator. I capture the X and Y waveforms into a file which is processed by some Octave (MatLab) code to produce the plots.
 

Attachments

  • FrankenTracer.jpg
    FrankenTracer.jpg
    152.6 KB · Views: 472
#428
lhquam said:
Below is the schematic for my "FrankenTracer" setup. For the LU1014D I use only one of the 24V transformer secondaries to get 24V RMS with 120V RMS into the primary. The Xout and Yout signals go to a Rigol DS1054Z Digital Sampling scope and the Vgs input to the SIT gate comes from a USB controlled Rigol DG1022A signal generator. I capture the X and Y waveforms into a file which is processed by some Octave (MatLab) code to produce the plots.
Click to expand...

That would be the equivalent of remote voltage sensing. A clever way to do it.
 
#432
woofertester said:
I was interested to know if the LU curves go concave down in the area where we would be cascoding these puppies. Looks all concave up at 3V up to 20A.
Click to expand...

Are these tests on the LU1014D's mounted on the aluminum IMS PCB TO247 adapters that I sent you? Are you using a fan to remove the 60w?

I was running my amp at 6A bias current today - no problem. I did use a fan on the CPU cooler though.
 
#433
xrk971 said:
Are these tests on the LU1014D's mounted on the aluminum IMS PCB TO247 adapters that I sent you? Are you using a fan to remove the 60w?

I was running my amp at 6A bias current today - no problem. I did use a fan on the CPU cooler though.
Click to expand...

Good question. This is a naked device with no heatsink. The test is relatively fast will small heat build up. I am awaiting the quick connect pins so that I can run the mounted devices that you assembled.
 
#437
LuDEF damping factor

My simulations show the LuDEF to have a damping factor over 100! Mind boggling for a non-feedback amplifier. This is due to the high transconductance of the LU1014D with no degeneration.
There is an ample supply of low hanging fruit to pick using the LU1014D. Get busy with designs.
 
#439
Zen Mod, transfering the discussion here from the post #14 in the "LuDEF kit packages" thread.

Zen Mod said:
first question - soldering small SMD critters is in fact lesser part of the job - what's most of the job is obtaining them , then matching them ( in batches of 50pcs, as I'm used) , without spilling entire quantity on the floor ....... :rofl:

so, 4 matched triplets - easy peasy to solder, but , as I said, boring and tedious to get there

LT3092 are bigger critters, so soldering them is even easier

in fact, I'm lying ..... I'm enjoying every part of it, every minute spent on preparing , matching ........ my brain is free to rummage and think of new/next things
Click to expand...

So, what is your answer regarding the sell of the PCB plus the matched but not-soldered SMD 🙂

Zen Mod said:
now, mounting - these pcbs I'm making lately, format is based on FW Format ;
most of them can be mounted either in regular FW way - using L or even better - T bracket, or they can be mount parallel to heatsink, say that you can see it as a whole in blog post: SissySIT, another one – Zen Mod Blog , while few more details you can see in post #3: Babelfish M25, SissySIT - general building tips and tricks

if you visit thread linked in first post here, you'll see enough details regarding tiny LU bugger, and why I resorted to mount it that way;

But!
That's also the reason why LuDEF channel pcbs can be mount only using L or T bracket

every construction , be it in electrical (schm) or physical (pcb) realm - is nothing more ( and less) than group of compromises; same here - I had to take care about physical arrangement of traces connecting LU, mounting arrangement, convenience of same for Greedy Boyz ( or anyone else assembling it) etc.

not saying that thought proces of making that all is hard - it isn't - it is in fact deeply engaging and enjoying in same time, but fact of that is not decreasing importance of every decision and level of overall complexity
Click to expand...

I know nothing about making PCBs, so I'm not arguing. And it's too late anyway as no doubt you have a bunch of PCBs make already.

But if I look at these PCBs, it looks for me it's not so complicated to adapt the right channel so that the LU would be rotated 180 degrees and mounted the same way as on the left channel (without a special hole). Especially if you leave out the jumper connecting the drain of LU to the area going to M103. I see the drain of LU is well connected to that area on the bottom layer. Vias next to the C107 marking could stay, connecting the two layers together. I think jumper is only there for better connecting the layers? Uhm??

I any case, I feel I have a good reason to learn using KiCad now 🙂
I would even try to make the PCBs mountable parallel to the heatsinks without the T-bracket in a way where Mosfets and LU are sticking out from the side of the PCB alowing perhaps to use some clamp as this for the LU 🙂

Anyway, in a very unlikely event of success, I would share it with you only 🙂

Sincerely,
-Alvis
 
#440
non- soldered JFets :
- including another variant/kit Option - complicating my already complicated life (I could include option without JFets, if one is having T092 as preferred option, but frankly, I forgot that; even if my opinion is that saving T092s for other project could be logical and clever)

- moving bottom located LU to top side - you need to crisscross both current traces - D and S; you also need to cross G trace, which is less complicated and less critical but all of them ........ unnecessary complication

learn KiCad and make them as your heart wish 🙂
 
You must log in or register to reply here.
Top Bottom