I don't know what's going on, but the prices on the small 500V/100V Ixys depletion mode devices have gone way up at the distributors. Maybe Ixys' new owner Littelfuse has decided to de-emphasize that product line...
The only problem I can see with the vacuum-type beasties is that their hungry, hungry filaments suck almost as much power as what you're trying to deliver to the speaker load.
Having said that., pentodes have their advantages. The jfets and mosfets kinda emulate pentodes with the flatness of drain current vs. drain voltage characteristic, but they still have that pesky Miller capacitance.
Having said that., pentodes have their advantages. The jfets and mosfets kinda emulate pentodes with the flatness of drain current vs. drain voltage characteristic, but they still have that pesky Miller capacitance.
Just got my PCBs today, and I have a heat sink picked out, that will get a little air flow from a throttled-down 48V 80mm fan.
Stuffed a fair number of the parts for the first version of the amp that will employ a Semisouth 1200V 85 milliohm fet or equivalent device from United Silicon Carbide. I may go for the USC device, as they claim to have improved the junction-to-case thermal resistance to a value more in line with Class A operation. I tested out a couple of my little 1kV IXYS depletion mode fets that will be going in the front end of this monstrosity and got a pair that bias up at around 2.5V reverse bias for 10 ma drain current. That's a little better than I remember than for the Supertex DN2540N5 devices, which have a lot higher gate capacitance, probably meaning they need more reverse bias for 10 mA of drain current.
Got a chance during the holidays to chop the heat sink in half,( once I figured out how to re-mount the non-ferrous blade back on my chop saw) spot, drill and tap the holes for the PCB supports, and do preliminary placement for the power semis (cap multiplier mosfet and SiC JFET output device). More drilling and tapping this weekend. I have some 4 mil mylar sheet on order to channel the air from the 48V fan through the heat sinks. It's a modest 80 mm fan with 0.08A draw at 48VDC - I'll be dialing it down to half that voltage for quieter operation, probably by using a couple of series zeners and a filter cap.
Here's a pic of the new amp channels sans output transformers. I figure I'll use either a pair of One Electron UBT-3s or a couple of vintage 3K Transcendars (full size) I have floating around. I plan to look at the output with and without global feedback. The HV bench supply I have at work will do to power up the channels for initial investigations.
Attachments
So am I - I just hope that the airflow will allow operation with the smallsh heat sinks I'm using. I plan some creative use of mylar sheet and kapton tape to channel the air where it will do the most good.
The Semisouth jfets will be up front next to the fan.
The Semisouth jfets will be up front next to the fan.
This past Saturday, I hooked one of the PCBS up to a One Electron UBT-3 output transformer and populated the components in the input Vbe multiplier. I'll see how it acts under power this weekend, and whether the Schade feedback loop needs some compensation.
Not going so far - this one is getting multiplexed with 3-4 other efforts, a full-time job, and a volunteer college radio DJ gig.
I wonder if anyone has a few of those 085's they would be willing to sell and if so
how much would they cost.
how much would they cost.
There is a silicon carbide JFET marketed by United Silicon Carbide/Qorvo that is a fair equivalent, with similar RDSon and a little more input capacitance than the Semisouth 085.I think both Digi-Key and the Mouse have them.
Edit - it's actually just the Mouse - this device has a little lower RDSon than the 085, and supposedly better juction-to-case thermal resistance.
https://www.mouser.com/ProductDetail/Qorvo/UJ3N120070K3S?qs=byeeYqUIh0OCiO%2B%2BGjwWnA==
Edit - it's actually just the Mouse - this device has a little lower RDSon than the 085, and supposedly better juction-to-case thermal resistance.
https://www.mouser.com/ProductDetail/Qorvo/UJ3N120070K3S?qs=byeeYqUIh0OCiO%2B%2BGjwWnA==
Last edited:
I had already looked at that device but didn't remember that that
the Simisouth device also had a rather high RDSon. It's a pity that
device has to have from -8 to -12 v on the gate for most of our designs.
the Simisouth device also had a rather high RDSon. It's a pity that
device has to have from -8 to -12 v on the gate for most of our designs.
Ugs and xconductance
first - ease of construction
second - resulting OLG, then all necessary choices
first - ease of construction
second - resulting OLG, then all necessary choices
Woody - the Semisouth device had an RDSon of only 85 milliohms for a 1200V device. An equivalent silicon device would be enormous, with a correspondingly high gate capacitance. The die was rather small for a device of that rating, and had about the same junction to case thermal resistance of something like an IRF130. That made thermal management challenging for the SiC Puppy amplifier.