Author Topic: 10 meter NPN bipolar finals are long obsolete, question changing to MOSFETs  (Read 737 times)

0 Members and 1 Guest are viewing this topic.

Offline floobydust

  • Frequent Contributor
  • **
  • Posts: 823
  • Country: ca
Working on an vintage CB radio and find NPN bipolar finals are long obsolete, Mitsubishi/Sanyo etc. discontinued the lineup about 2004. Example was 2SC1969.  eBay and scalpers sell some, but mostly fakes or silly expensive.

Many are using IRF520 despite it's high input capacitance Ciss 360pF. The IRF510 is half that, at 180pF.

I'm not sure about the MOSFET bias circuit.

Paul Harden NA5N has a very good article comparing MOSFETS to bipolars in QRP power amplifiers.
It's what I thought - a trimpot to set bias around VGS(th) at 2-4VDC, just out of Class C a tiny bit.

But EKL Components Palomar offers a bipolar->MOSFET replacement ERF-2030 now ERF2030+ (no datasheets) with companion thick-film DC bias networks, like EN369DR. Needs a pull-up resistor to 9V TX I believe.
They have an app note for Magnum 10 meter transceivers, and a few other radios.

I don't get why their bias module has a (reversed) 1N4148 diode and what it is trying to accomplish. I thought you want a symmetrical load on the driver stage.
Any insights appreciated.

edit: updated ERF part number/vendor
« Last Edit: November 14, 2017, 07:37:04 AM by floobydust »
 

Offline xaxaxa

  • Contributor
  • Posts: 45
  • Country: ca
You may want to use RF LDMOS transistors as those are specifically designed to be rugged against temporary overvoltage from high vswr.

IRF510 has a reputation of blowing very easily when used as a rf amplifier.

The mrf184 might do the job; it is an unmatched 900MHz transistor so can operate anywhere below 900MHz. The input capacitance is 83pF. Other candidates include mrfe6s9125n, mrfe6s9160h, and mrf5s9070n. These ones have an internal matching network for 900MHz but at 30MHz it is invisible (the series inductor looks like a short and the shunt capacitor looks open).

The gate to source diode conducts when the gate voltage goes negative and charges up the dc blocking capacitor, and gives you an increased bias point as the drive is increased. It isn't strictly necessary.

 

Offline floobydust

  • Frequent Contributor
  • **
  • Posts: 823
  • Country: ca
I'm not sure why IRF510's are said to be unreliable in RF PA.
Looking at VSWR, bipolars are VCEO 25-30V and the IRF510 100V; Mitsubishi RD16HHF1 50V MOSFETs said to withstand 20:1 VSWR  Po=16W f=30MHz. Unless it's parasitic oscillations or hot spots that kill them.

I did notice Ranger Galaxy uses IRF520's and 5.6V zener to soft-clamp gate voltage (through 1k resistor to gate) on all their MOSFET stages. It looks like a good idea to protect gate voltage this way.

The Palomar ERF2030+ (EN1230) bias circuit I ran through LTSpice and it gives about an extra volt of bias but fixed-bias seems best.
Those 900MHz parts are pricy around $48 and I need the TO-220 package.

I'll start with IRF510 or 520 and see how they do. They are around $1 each.

The hardest part is the driver - it's designed for a bipolar (final) so not sure how it will do for output voltage vs MOSFET input capacitance.
 

Offline DC1MC

  • Contributor
  • Posts: 31
  • Country: de
Why don't you use the drop-in replacements NTE236 http://www.nteinc.com/specs/200to299/pdf/nte236.pdf

Available for a few bucks everywhere http://www.newark.com/nte-electronics/nte236/rf-bjt-npn-25v-6a-27mhz-to-220ab/dp/31C3573, I doubt that in the exiting circuit the FETs will work better and be less expensive.
 

Offline CJay

  • Super Contributor
  • ***
  • Posts: 2067
  • Country: gb
  • 2E0EOA
Look up Eleflow, they supply drop in bipolar replacements for the popular Mitsubishi and other transistors that were used in CB gear as well as a lot of other HF gear of that era.

If you really want to convert to MOSFET then there are details of the modifications you need to do on a few CB websites, CBTricks, DefPom (though he charges for access to some info or did)

There are still some bipolar devices available, 2SC2078 is one that pops up in new gear so someone is making it but eBay is rotten with fakes.

It's worth having a look for the schematics for 'Export' radios, they're almost all using IRF520 devices or the FQP13N10.

I've experimented with the Fairchild part, it's capable of insane power, 52W running on 25V at 28 ish MHz, but it's also incredibly delicate at those power levels, I let the smoke out of a few while chasing the limits but if you treat them kindly they're pretty good.
2E0EOA
 

Offline floobydust

  • Frequent Contributor
  • **
  • Posts: 823
  • Country: ca
NTE236 or 2SC1969 is only rated to VCEO 25V, seems low and no room for poor SWR? But datasheet says "...ability to withstand infinite VSWR when operated at 16V, Po=20W, f=27MHz."

The stock Sanyo 2SC2078E 2SC2092 is 75V 3A 45-60pF, hFE 100-200, fT 150MHz, crosses to NTE235 but hFE range is 25-200; and it's around $10+shipping. Eleflow 2SC2078 is 35V.

I thought if I have to mess around with tuning the output capacitance and base drive on a substitute part, why not change to a $1 IRF510 or IRF520 and at 100V it might be a bit stronger for poor SWR.

No end of terrible CB hacks out there to get more than the legal 4W AM. I didn't realize until I saw the bonehead youtube videos  :palm:
Not my intent, I'll try changing to the IRF's as a learning experience. I wish I could check IMD or access a radio test set.

Just waiting to get the receive section working, and for $3.00  50R 250W dummy load resistor to arrive and assemble a heatsink, current transformer, as my old #47 lamp is not the best ;)
« Last Edit: November 30, 2017, 03:56:51 PM by floobydust »
 

Offline CJay

  • Super Contributor
  • ***
  • Posts: 2067
  • Country: gb
  • 2E0EOA
Which radio is it you're working on?

There's lots of nasty out there, so many idiots who know nothing other than how to screw up radios but for some reason they always have clients, I spent a lot of time removing and cleaning up 'experts' work but finally gave up after having fixed one too many radios that had been expertly buggered up by one of those screwdriver wizards.

The FQP experiments, I was putting them in a homebrew ampifier and for the sheer fun of it I was seeing how far they could be pushed on a dummy load before the smoke escaped, they're used in a lot of export radios in pairs to good effect and are so cheap I decided to play a little.
2E0EOA
 

Offline AF6LJ

  • Supporter
  • ****
  • Posts: 2577
  • Country: us
I think if you take the time to look around you can find a bipolar replacement. A 2SC1969 isn't all that special.
Sue AF6LJ
Test Equipment Addict, And Proud Of It.
 

Offline floobydust

  • Frequent Contributor
  • **
  • Posts: 823
  • Country: ca
This is a dumpster rescue, early 1977 cobra 29XLR lying around my junk collection for many years.

The finals had shorted and trucker Joe had no fuse it seems, melted the modulator transformer and a bunch of other inductors. So I replaced all that and it was transmitting somewhat with an old junkbox RF PA transistor but it died. It's $12 for another or $1 and some fun.

I drew out a schematic of the original transmit section and proposed changes, three bias possibilities.
Not sure if it's better to limit gate drive, or output tuning, to keep power proper. If transconductance varies with temperature, might it be moody then.
 

Offline CJay

  • Super Contributor
  • ***
  • Posts: 2067
  • Country: gb
  • 2E0EOA
Honestly wouldn't bother with the MOSFET conversion, it's a standard 40CH AM radio which will work just fine with any number of transistors and need absolutely no alteration to suit them, 2SC2078, 2SC2092, 2SC1909, 2sc2029, 2sc1678 and any number of others should be just fine.

The Cobra 29s have a good reputation of being one of the best 'simple' radios built but they're still engineered to a price, it really wasn't uncommmon to be able to open five different factory fresh radios and find two or three different final transistors with no other circuit differences, some manufacturers even listed the alternatives on the schematic...

if you want to tinker and have some fun though, grab a pack of the FQP13N10 devices and work out some biasing, be prepared for some foul smelling smoke as they and the IRF devices will self destruct in a fraction of a second if you get it wrong.

I have a few finals here which I've harvested from various radios, happy to stick one in a padded envelope and send it as an early Christmas present if you want one?
2E0EOA
 

Offline AF6LJ

  • Supporter
  • ****
  • Posts: 2577
  • Country: us
This is a dumpster rescue, early 1977 cobra 29XLR lying around my junk collection for many years.

The finals had shorted and trucker Joe had no fuse it seems, melted the modulator transformer and a bunch of other inductors. So I replaced all that and it was transmitting somewhat with an old junkbox RF PA transistor but it died. It's $12 for another or $1 and some fun.

I drew out a schematic of the original transmit section and proposed changes, three bias possibilities.
Not sure if it's better to limit gate drive, or output tuning, to keep power proper. If transconductance varies with temperature, might it be moody then.
Whatever you do have fun, do watch the gate drive. especially since in many of those old AM radios the driver and final are modulated and those gate drive voltage peaks can get you.
Sue AF6LJ
Test Equipment Addict, And Proud Of It.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf