ERF2030+ new version indestructible?

[Robb]

One thing I just learned was that the input and output capacitances of these two parts is far different than I thought and in the opposite direction. For the 13N10 the Ciss was 567pf and Coss was 558pf. The ERF2030+ came in at a much higher Ciss of 1422pf and a Coss of 1270pf. That is enough to make a difference in the tuning of circuits and confuses me because no changes were made in the PA stage other than the transistors.

No bending of coils and no changing of any bias gate voltage. The only thing I can think of now that would be making the ERF2030+ work better in the Stryker SR-94HPC is that the transistor must have a faster rise and fall time where the 13N10 is at its limits.

The current reduction has made a noticeable difference in how cool the TIP31 series pass modulator runs too. The cast aluminum housing takes much longer to heat up and never gets as hot as it did. If you're concerned with how hot this radio normally runs when you get long winded, don't bother to change the modulator transistor, change both finals to the ERF2030+.

This radio seems to lack a tuned matching circuit between the driver and final stage like most CB's and looks capacitively coupled. That may explain why it handled the change in higher input capacitance over the stock finals. It doesn't explain why with no output impedance adjustment I saw and increase in power and a reduction in current, other than perhaps the original match was off.

I would experiment more with the output match if I had a schematic. I don't want to get into a situation where I can't identify output matching from things like 54 MHz. traps and the radio is working well so I think I'll stop fixing it before I break it again.

This radio may not even use any DC voltage on the gates. I didn't measure it but they may just be driving the finals with enough power to run them with zero bias. It did use one driver the same size as each final and that's a lot of drive capability. Think of your typical "3 pill".

The theory of zero bias being used here eliminates the pesky variable of riding the gate threshold voltage perfectly to maintain the gain required to produce rated power output. It's not so fussy if you have enough drive to eliminate the need of DC to turn the transistor on. Would be nice if they started publishing some idling current specifications for various classes of bias on these parts...

Then we would know right away if lack of power output was related to bias problems or matching problems on the input or output. Sad to say, Palomar can't even seem to provide any important specs like the gate threshold voltage for comparison so each parameter has to be tested on your own to find out as needed.

In most cases, the higher output capacitance of the ERF2030+ should require changes in the output matching such as a reduction in the value of any external source capacitor and bending of the next coil.

Are you planning to tweak the circuit so that maximal and reliable estimates can be known, as well as other specs?
Thnx!

 

[Shockwave]

Are you planning to tweak the circuit so that maximal and reliable estimates can be known, as well as other specs?
Thnx!

Since it's making more power and drawing less current, I'm inclined to leave it alone without any alignment data. I'm sure I could probably get it to do more in one spot if I tried but it's fairly flat across the bandwidth now and peaking it on one frequency without the service info is likely to cause roll off at the bandwidth edges. Both finals and the series pass modulator are running cooler so reliability had to go up too.

At 52 watts from a pair of parts rated at 30 watts maximum PEP each, that's even a bit more than I'm comfortable with...

 

[Shockwave]

Hmmm...



I think it's been said before...

https://www.worldwidedx.com/threads/palomar-erf-2030.233381/page-6#post-735528

Only just before the poke in the eye with the sharp stick...

But then who am I?


Just a duck...

In that schematic they are driving the finals through a coupling cap that is 5.27 times lager in value than the internal gate capacitance of both final transistors in parallel. That gets the vast majority of drive, into the low impedance of the gates. Two of those 152 caps in parallel to make it 10 times the value and half of the inductance might bring it up more. At this point the cap is considered a short to RF.

The output side is where less parallel capacitance or more series inductance might improve the match by compensating for the larger source to drain, internal capacitance.

 

[Handy Andy]

But here's the rub, this is what Magnum, (RF Limited) and Stryker idealized to. So your drop in, the ERF2030+ is using those values.

The RFX-xx series use a similar acceptance input capacitance to even drive their second stage...and that's even an older approach to this problem.

They also "offset" or parallel a simple resistor (R216A) on the Final Stage to divide the EN part used, divides the power nearly in two for the swing in drive levels the Finals will see. Even though they already have a R216 to act as a cushion.

When they Had ERF2030's - not all of them needed it.

Now in some chassis they don't even use these - but they do put in 13N10's - and we've seen how that goes...

Poof...

So in some ways, they scrambled to fit parts in there to replace that which they lost. In doing so, earned a reputation - but also learned how to compensate for the different installs - we just had to pay attention to the parts NOT used when the parts they use are different - they are subbed in at the factory for other reasons.

So to me, the ERF2030+ is the HIGHER Gate capacitance part they were looking for, so it looks to me that they kept the larger values in there for the sake of simplicity for upgrades - you, the tech, then had to determine the changes needed in offsets to handle the changes in needed the power drive levels to replace them.

It's the only way to see how these parts they wanted, but not installed, even get mentioned - to help the rest of us find the correct levels of capacitance to drive, and offset, using resistance, to divide down the power to make it work.

At least that's how I see it, those Bread Crumbs they left behind.

 

[Shockwave]

But here's the rub, this is what Magnum, (RF Limited) and Stryker idealized to. So your drop in, the ERF2030+ is using those values.

The RFX-xx series use a similar acceptance input capacitance to even drive their second stage...and that's even an older approach to this problem.

They also "offset" or parallel a simple resistor (R216A) on the Final Stage to divide the EN part used, divides the power nearly in two for the swing in drive levels the Finals will see. Even though they already have a R216 to act as a cushion.

When they Had ERF2030's - not all of them needed it.

Now in some chassis they don't even use these - but they do put in 13N10's - and we've seen how that goes...

Poof...

So in some ways, they scrambled to fit parts in there to replace that which they lost. In doing so, earned a reputation - but also learned how to compensate for the different installs - we just had to pay attention to the parts NOT used when the parts they use are different - they are subbed in at the factory for other reasons.

So to me, the ERF2030+ is the HIGHER Gate capacitance part they were looking for, so it looks to me that they kept the larger values in there for the sake of simplicity for upgrades - you, the tech, then had to determine the changes needed in offsets to handle the changes in needed the power drive levels to replace them.

It's the only way to see how these parts they wanted, but not installed, even get mentioned - to help the rest of us find the correct levels of capacitance to drive, and offset, using resistance, to divide down the power to make it work.

At least that's how I see it, those Bread Crumbs they left behind.

Good info Handyman. I see the SR-94HPC is using a discrete surface mount 1N4148 diode and resistor to make the EN part too. I'm starting to think anytime we see the EN circuit used, it's because a switch mode transistor needed bias it can't handle continuously.

Like in the AM mode where carrier and bias would both be present on an FET. This part reduces bias on carrier and increases it on word peaks to provide enough gain to produce swing. Not enough bias for SSB but about all it can handle with carrier on AM. No other real linear RF transistor uses this Mickey Mouse variable swinging bias scheme. Notice on SSB they apply a larger fixed bias voltage through another transistor.

I'm sure you see these things but just wanted to point them out for those wanting more details.

 

[Shockwave]

Another thing I'll mention is this coupling of the driver to the final stage using just a large value cap, worked great with bipolar transistors because their output impedance is lower than the input impedance. Making it very easy to drive the next stage without any additional matching required.

Most MOSFET RF power transistors are the exact opposite. The extra gate capacitance causes the input impedance to be lower than the output. That means even if you use a cap with a value high enough to look like a short to RF, you still fail to get all of the power from the higher output impedance driver, into the lower impedance at the gate of the finals.

To accomplish this would require additional matching components. CB rigs skip that in favor of just using the same driver as the final. That gives them extra drive power to waste in the mismatch between driver and final.

 

[Shockwave]

One thing I just checked on the Stryker SR-94HPC, was for any type of DC shunt across the antenna connector. Knowing that the final blew up when switching from an antenna that is DC grounded, to one that is not, made me question if the radio lacked the ability to drain any static charge off the antenna line.

How about a DC resistance of 10K? While that is low enough to provide a consistent load on an antenna to prevent static buildup, it is not low enough to prevent damage if a DC imbalance or static buildup is instantly switched onto the radio.

Coincidentally, this radio also blew open its front end FET about a year ago. While I was not alert enough to actually notice when the radio went deaf, like this time when the antenna switch box was switched, it is likely the antenna that is not DC grounded, contributed to the front end receiver failure as well as the final that recently failed.

That's enough to make me install a 10uh DC choke across the antenna connector. Other radios with similar FET final stages and front ends could benefit from adding this part too. They may be able to handle overload and poor VSWR better than expected but, may be suffering from failure due to switching transients at the antenna connector.