What have I done wrong (Mosfet Conversion)

[TheRealPorkchop]

I don't know why I'm not receiving notifications of replies... I believe I found what you pointed me to, Sonoma, except for dual finals. Much thanks as always, brother.

Nomad, I don't know so much about them as I'm sure you do. But I do know that you can't get them anymore and building them just blows. I think it makes more sense to convert to mosfets and use a factory setup for adjusting bias, I can't speak for anything SSB related as I don't ever fool with it. Well, I've talked sideband but I don't ever test anything on sideband and all my SSB conversations are held on a Yaesu.

This is what I found and this makes more sense to me :

http://www.cbtricks.com/miscellaneous/fet_papers/mosfet_conv/index.htm

But honestly I've looked into the Eleflow transistors and I'll probably give those a shot before I continue to fool with mosfet conversions, it's just annoying to me. I've got a 14 year old S9 with 1969's in and that have never been touched and I've ran that radio in my truck for the last 14 years. I feel like at some point, they'll bite the dust and I'll two choices, mosfet or original replacement. I wish I'd got some 2312's when I could and Eleflow is cheaper than RFP for 1969's ( I know they aren't the exact same piece.)

 

[sonoma]

never did understand why Rick and Bennie did not do a double mosfet upgrade for the AM only radios but the am ssb mod should get you in the ball park for the mod.

 

[TheRealPorkchop]

Well the radio in question is a Galaxy DX77, so it’s already dual final but I have a metric ton of mosfets on hand and zero 1969’s or 2312’s. Actually I have some 1969’s but I question their authenticity.

Using my Peak Atlas, the 2312’s have a whopping gain of 8 and 9. And the PEP watt reading is 30w. Seems ok, I know but I had one of these personally that I ran for years and got much more from it. So I figured, hey, the finals are weak and testing them made me raise my eyebrow to it. So digging through the bin after having my feelings hurt on dropping in some N10’s or 520’s, I ran across some 1969’s.

Now...the gain reading on one is 37 and the other is 43 but for doo-doo’s and giggles I figured why not. Dropping them in yields the same PEP reading and doing the infamous Red Wire mod gives me about 45w PEP. Something just ain’t right, I honestly feel like it should do that without volting the finals. But whatever.

So...to add one more clunker to my personal collection, I’m trying to restore this radio and put it like I had mine way back then. So mosfet modding is probably my next wasted time step. I guess I’ll just copy the final section out of a 33HP or something and see what happens. It’s not really a big deal and absolutely not important what I end up with. It’s really just in the name of experimenting and killing spare time.

 

[sonoma]

Porkchop you should be able to look at the ssb 3600 board mod and compare it to the radio you have and figure out what needs to be changed. .I assume your radio has the 3600 board in it. at least you have the basic ground work done by looking at the dual final mod.
one thing Rick told me a few years ago that the thick porcelin insulators work best due to a capacitance issue on the dual ,final mod to mosfets. . I did the mod on a tr696 mobile radio a few years ago and with the dual finals in it would swing to just over 20 watts and stop. changed to the porcelin insulators and it started swinging to over 40. with the mylar insulators in it you could hit 20 watts very fast and it looked like the needle on the meter was hitting a stop.

 

[TheRealPorkchop]

Oh yea, I won’t have trouble swapping it over, have more trouble sourcing the parts. Maybe not sourcing but acquiring them I guess I should say.

It has the ceramic insulators and I ordered a handful of several different sizes for such projects. I have the silicone ones too, less messy but not as cool looking.

 

[Handy Andy]

Well, if you're willing to wade thru this...​

Below is simplified, but if are trying to convert, you need something to reference anyways...

Ok, for helpers ...

Your divider? Galaxy kinda keeps them pretty high-values for VERY low current and even less for recovery...



So, Bias resistors? You see in the graphic the top one shows what you'd have for a typical Bi-polar, the hi-power low ohmic value resistor to drop the 8 volts down and they use a 100 ohm trim pot and a low ohmic value resistor paralleled with it to lower/divide the current draw between the two (1/2 each sort of thing) and provide the current oomph for the bias diode.

The bottom one? Pretty much used by Galaxy (done up to protect their interests) - but here's what I'd do for your 77...the biggest hint it to pay attention to the resistor values below in text ... they can be subbed in for above...

PAY ATTENTION!
READ THRU INSTRUCTIONS FIRST
KNOW WHAT YOU ARE DOING!
​

Use 1 watt 7.5K to drop from your 8 volt line and use IRF520 - use 4.7K from your drop line - to ground (your divider) then use a 1K on the ferrite bead to feed the IRF520's gate - before you hook up the gate, double check the voltage going to the gate - should not be higher that 3.8 volts - otherwise a possible latching issue. ( use 8.2K if you have to, on your 8 volt side to lower voltage - err on the side of caution here...)

To lower the voltage if you don't get the 3.8 - THERE IS HOPE - use a diode in-line with your 1 watt 7.5K or 8.2K - Banded end towards the rear panel - in SERIES with that 8 volt feed resistor...should bring you to between 3.7 to 3.4 volts - be done! (The use of the diode; as a watt-less dropper to trim your bias - else 3.9K on ground side of your divider is your only other option...)

For the 13N10s'? Use 8.2K 1 watt on the 8 volt line - and use 4.7K on your drop line (divider to ground side) use a 1K on a ferrite bead and check your gate voltage before you hook up - can't be any higher than 3.3V or risk of latching - to lower, use the diode trick on your 8.2K in-line with it banded end towards your rear panel. (if the diode can't do enough - use 3.9K as your drop to ground side to lower voltage even more...)

The Zeners don't help with swing - they tend to clamp and regulate the gate too quickly - but they do provide a level of protection - to use them is up to you - but most of the work will be done at the divider and the buffer resistor keeps current from "zapping" the gate and forcing a latching on.

Zeners also tend to "latch on" and suck down a lot of RF - to make them recover quicker they reduced capacitance filtering and made the resistor values so high to provide a recovery method for when the zeners latch.

13N10's and IRF520's can use some capacitance - so 560pF (561) on up to 102 (1000pF) to drive them...

Not responsible for error - for Educational Purposes Only
User of this information assumes all risks and liabilities...

Have fun...
:+> Andy <+:

 

[sonoma]

was hoping Andy would get in here to do a drawing for the mod.
thanks Andy, I copied this off also. not sure if I will ever use it but I have it for just in case. I have a galaxy 33 and 44 I may do the mod if they ever go down.
thanks Andy
Sonoma

 

[TheRealPorkchop]

Nice. Thanks Andy, you’re the man!

So basically with your drawing and what you’re saying is, you could do this mosfet conversion and only change those resistors to use the VR’s in it from the factory to control voltage to the gate? The only reason I haven’t just followed one of the other methods is a lack of parts. I always seem to have half or none of all the required parts. When I order my next batch of parts, I’m gonna get more than needed.

 

[Handy Andy]

Remember the VR values, the old BI-POLAR ones are setup for something much much lower ohmic values...The graphic is to show you what Galaxy did for their DX MOSFET "conversion" using Bi-polar as the comparison...

Later on...they did the 10X values on Driver (the 47K - 100K values) then use 1X values for FINALS (4.7K and 10K)...IRF520 stuff - so they adjusted the bias values by factors of 10...

The posts' text has what you need to do - in values - pots are removed - everything is fixed now.

I'll go back up and edit that post, the resistor values change, you don't have to use the pots anymore...

Why? Well...Take a look at the 687 - from Uniden - they use a IRF520 but their "resistor divider is pre-set" and uses no diode...


The thing also needs a bit more work and the inductive output tank is part of this problem...

The only bane is the fixed 8 volt section...

The Diode part is my work, the dividers aren't perfect for every case - so you'll need to trim the drop - and I usually use the lower side (referring to voltage - LESS GATE VOLTAGE) due to the thermal effects when the Gate starts to heat up - they can latch at voltages you set originally and when they warm up? POW-POOF-POP (you get the idea) because the Gate voltage "turn-on" threshold DOES CHANGE as temperature goes up...

So...

- applying the diode on the SOURCE side - DROPS DOWN your voltage across the divider...
- Applying the diode on the GROUND side - RAISES your voltage across the divider...

Applying means IN SERIES - with the power flow - (Banded end directed to ground) with the resistors involved, not parallel to them - The diode will have inherent intrinsic impedance against the power flow in the divider circuit. So this affects the divider in a fashion that the diode can provide a trim function to either raise or lower gate voltage as you need to tune the section....

The raise and lower of voltage reference means - It is measured at the GATE lead of the device before you install. The Ferrite Bead and Resistor are USED - so they need to be installed BEFORE you test, for GATE voltages to be correct

- You can use fixed resistors of lower values - will raise current thru the GATE which can help in some instances for drive. HOWEVER, err of the side of caution when it comes to applying too much current - it will affect SSB drive. IT is why I recommend the 1K and Ferrite bead combos'

- It is far safer to use 8.2K or 7.5K for all sourcing mods for FINALS
- and for the GROUND side of your divider use 3.9K for 13N10s' and 4.7K for IRF520s' - CHECK GATE VOLTAGE before attaching the Gate VERIFY your work.
- Apply Diode mod's as needed - but remember to treat the diode like a BIAS diode - conventional current thinking - power flows to GROUND - SO AIM the BANDED end towards ground.​

Be careful out there ok?

:+> Andy <+: