But the function also calls?setFrequency(), which does call the Si5351 routines. The stability of the entire process seems to depend on the analogRead() that sets knob. Given the 10-bit resolution, there's room for some slop in the mapping process. Note how the analogRead() value is expected not to change at all between calls to doTuning() if the user is not changing the pot. The fact that I see -10 after every read tells me there's some slop in the measure. Personally, I'd feel better if we tested the knob using a range (+ or - X) before we say the user has actually changed the frequency.?

If the jitter is constantly moving, we might think it is VFO drift. Your description makes it sound like it's jumping around a central frequency that really isn't changing.

Jack, W8TEE

---

From: Joel Caulkins <caulktel@...>
To: [email protected]
Sent: Thursday, February 16, 2017 3:23 PM
Subject: Re: [BITX20] KB6QVI: VFO

Jack,

As the last digit jitters the frequency jitters also, so it's not the display I'm worried about. It sounds like the person is gargling:-)

Joel?

KB6QVI

On Feb 16, 2017, at 12:06 PM, Jack Purdum via Groups.Io <econjack@...> wrote:

Joel:

My B40 is apart, but try these while you wait:

void doTuning(){

?unsigned long newFreq; ? ? ? ?// Not used...remove??

?

?int knob = analogRead(ANALOG_TUNING)-10;

?unsigned long old_freq = frequency; ? ? ? ? ? ? ? ? ? ?// Not used...remove??

if (knob == old_knob) ? ? ? ? ? ? ? ?// New...add

? ? return;

? // the knob is fully on the low end, move down by 10 Khz and wait for 200 msec

?if (knob < 10 && frequency > LOWEST_FREQ) {

? ? ? baseTune = baseTune - 10000L;

? ? ? frequency = baseTune;

? ? ? updateDisplay();

? ? ? setFrequency(frequency);

? ? ? delay(200);

? }?

? // the knob is full on the high end, move up by 10 Khz and wait for 200 msec

? else if (knob > 1010 && frequency < HIGHEST_FREQ) {

? ? ?baseTune = baseTune + 10000L;?

? ? ?frequency = baseTune + 50000L;

? ? ?setFrequency(frequency);

? ? ?updateDisplay();

? ? ?delay(200);

? }

? // the tuning knob is at neither extremities, tune the signals as usual

? else if (knob != old_knob){

? ? ?frequency = baseTune + (50L * knob);

? ? ?old_knob = knob;

? ? ?setFrequency(frequency);

? ? ?updateDisplay();

? }

}

I have no idea if this will work, but essentially is says that if we haven't move the tuning knob, don't bother updating the display. Let me know what happens.

Jack, W8TEE

---

From: Joel Caulkins <caulktel@...>
To: [email protected]
Sent: Thursday, February 16, 2017 2:32 PM
Subject: [BITX20] KB6QVI: VFO

I just ordered one of these to use on my BitX40, the Raduino just isn't working out for me, too much last digit jitter.?

Joel,?KB6QVI

Though Raj's resistor in the source lead could prove to be the best solution, gives an instantaneous response to excessive current.? The G0KCL circuit is slower (due to filter caps) and the PolySwitch much slower (has to heat up), but either might be adequate to prevent failure here in most cases.? I hope to try some of these schemes out someday soon, safety goggles on and a stash of spare IRF510's at hand. ?

Maybe we need to write up a hack to move this to an RD16HHF1 for those so inclined.? Perhaps a pair of them.? More robust and better behaved than the IRF510.? Popular with those willing to spend $5us on a transistor.

On Thu, Feb 16, 2017 at 10:31 am, Jerry Gaffke wrote:

The G0KLA circuit might be a better bet, it only drops out by around 0.6 or 0.7 volts. ?

?

Looking at the LM338, a 5 amp regulator, the app notes show a constant current regulator circuit.? It should work with a LM350, the 3 amp version.? This might be the easier solution to restrict the current to the IRF510.? It also could be created so that it just plugs into the PA power port.? It will need to be tested, to see if the output voltage varies much from the input voltage.

Marco - KG5PRT

On Feb 16, 2017 11:25 AM, "Jerry Gaffke via Groups.Io" <jgaffke=[email protected]> wrote:

A fuse could be good enough to save the IRF510 in many cases, if it is being destroyed by heat rather than avalanche mode.? An alternative to a fuse might be a PolySwitch such as this: ?MINISMDC110F/24-2. ??It resets itself once it cools down, no need to replace it.?

> ? Raj vu2zap wrote: ? A fraction of an ohm resistor in the source lead?would provide a negative bias with increasing device current. There will be a small drop in output power I guess. The resister can be bypassed with a cap.

Well worth trying.? Doesn't take much change in Vgs to significantly cut back on drain current, as anybody who has twiddled VR1 well knows.? Wouldn't waste any more power than the shunt resistor that G0KLA uses to sense the drain current with his scheme.

For those figuring out blown IRF510's, or just puzzled about how that final amplifier works, here's a very quick tour:

L6,L7 form a low-pass filter, it's symmetrical so should be 50 ohms straight through at 7mhz.? C155,C156 are dc blocking caps to keep 12v off of the antenna.? The 2-to-1 winding ratio at T7 transforms that 50 ohms to 200 ohms as seen by the IRF510 drain. ? C153,C154,C157,L8 filter the incoming 12v power supply.? The RV1 gate bias of roughly 4 volts comes in through R150.? The incoming 7mhz signal comes in through the dc blocking cap C152 from the previous stage, driven from the same 2-to-1 winding ratio transformer as the IRF510 has. ? Current down through the IRF510 from drain to source is controlled by the gate voltage.? The gate is well insulated from both drain and source, but capacitance there (mostly from the Miller effect) can suck a lot of power from the previous stage when running at 7mhz. ? Simple enough.? But there are many ways this can all go wrong.

Figure 1 on page 3 of this datasheet gives some idea of how drain current varies with gate voltage: ?? ? But the Vgs(th) spec of 2 to 4 volts on page 2 suggests the gate voltages shown in figure 1 might vary by up to a couple volts between different IRF510's to give the same current.?

Regarding John Smith's constant current regulator: ?We would like a supply that gives a constant voltage, but only up to some maximum current.? There are two ways this can be implemented.? Could simply limit the current to that maximum level, allowing the voltage to drop as it may, John's "constant current".? Or could drop the voltage to near zero when the current limit is exceeded, rather like a circuit breaker, until the operator resets it somehow.? The latter case is called "foldback" current limiting.? Not many regulator IC's bother with providing any kind of easily configurable current limit, which is unfortunate.

A fuse could be good enough to save the IRF510 in many cases, if it is being destroyed by heat rather than avalanche mode.? An alternative to a fuse might be a PolySwitch such as this: ?MINISMDC110F/24-2. ??It resets itself once it cools down, no need to replace it.?

> ? Raj vu2zap wrote: ? A fraction of an ohm resistor in the source lead?would provide a negative bias with increasing device current. There will be a small drop in output power I guess. The resister can be bypassed with a cap.

Well worth trying.? Doesn't take much change in Vgs to significantly cut back on drain current, as anybody who has twiddled VR1 well knows.? Wouldn't waste any more power than the shunt resistor that G0KLA uses to sense the drain current with his scheme.

For those figuring out blown IRF510's, or just puzzled about how that final amplifier works, here's a very quick tour:

L6,L7 form a low-pass filter, it's symmetrical so should be 50 ohms straight through at 7mhz.? C155,C156 are dc blocking caps to keep 12v off of the antenna.? The 2-to-1 winding ratio at T7 transforms that 50 ohms to 200 ohms as seen by the IRF510 drain. ? C153,C154,C157,L8 filter the incoming 12v power supply.? The RV1 gate bias of roughly 4 volts comes in through R150.? The incoming 7mhz signal comes in through the dc blocking cap C152 from the previous stage, driven from the same 2-to-1 winding ratio transformer as the IRF510 has. ? Current down through the IRF510 from drain to source is controlled by the gate voltage.? The gate is well insulated from both drain and source, but capacitance there (mostly from the Miller effect) can suck a lot of power from the previous stage when running at 7mhz. ? Simple enough.? But there are many ways this can all go wrong.

Figure 1 on page 3 of this datasheet gives some idea of how drain current varies with gate voltage: ?? ? But the Vgs(th) spec of 2 to 4 volts on page 2 suggests the gate voltages shown in figure 1 might vary by up to a couple volts between different IRF510's to give the same current.?

Regarding John Smith's constant current regulator: ?We would like a supply that gives a constant voltage, but only up to some maximum current.? There are two ways this can be implemented.? Could simply limit the current to that maximum level, allowing the voltage to drop as it may, John's "constant current".? Or could drop the voltage to near zero when the current limit is exceeded, rather like a circuit breaker, until the operator resets it somehow.? The latter case is called "foldback" current limiting.? Not many regulator IC's bother with providing any kind of easily configurable current limit, which is unfortunate.

Thank you Ron for your help, unfortunately desoldering dual diode and replaces two 1N4148 plus 100 ohm trimpot not solve the problem ..... I am looking for further solutions .....

Tom

Perfectly said. I second Eddie Statement.? Thanks very much

Ric KM4LEN