Pages: [1]   Go Down

Author Topic: 813 parasitic oscillation  (Read 304 times)

K0IZ

  • Member
  • Posts: 1176
813 parasitic oscillation
« on: July 28, 2021, 06:05:00 PM »

I have my 20M 813 (2) grounded grid amp working.  Trying first without parasitic suppressors.  Static plate current is 40 Ma, plate voltage about 2300.  Without any driver load or plate output load, the plate current will jump around, up to about 125 ma, as I adjust the plate tune and load variables.  I'm assuming parasitic oscillation.

If I attach a 50 ohm load to the output, no changes whatsoever in plate current with tuning.  If I remove load and attach a 50 ohm dummy load to input instead, the plate fluctuations are about half as much.

I think I would prefer to not include parasitic suppressors.  So my question:  Assuming the amp will always have some sort of load, am I good to go?
Logged

K1KP

  • Member
  • Posts: 144
Re: 813 parasitic oscillation
« Reply #1 on: July 29, 2021, 06:03:58 AM »

Others with more experience in building tube amps may comment, but it seems to me that doing this testing with no input or output load is unrealistic. Under these conditions the amp has tremendous voltage gain, which means that a very small amount of coupling from output to input would result in oscillations. They would be near the operating frequency and not VHF. Adding realistic loads to the input and output is definitely a fairer test.

My opinion is you are good to go.

-Tony, K1KP
Logged

K0IZ

  • Member
  • Posts: 1176
Re: 813 parasitic oscillation
« Reply #2 on: July 29, 2021, 06:22:00 AM »

Good thought about the oscillation being at operating freq.  My prior experience with parasitic oscillations was that they were very erratic with tuning and severe.  These plate current changes are a little bit more subdued and not very sharp.

I was actually checking the bias and resting current, and didn't have input or output loads on.  Then noticed that sometimes the resting current was higher than other times.  Was 38 Ma sometimes (my target was 40 Ma), other times 75 Ma, etc.  Then I turned the tuning variable and saw it vary from 38 to 125 Ma. Once I saw that I knew I had some sort of oscillation.  Fortunately was for much less than a second and apparently didn't damage anything.

Adding load and it settled down very nicely without a flicker of plate current change.

I suppose I inadvertently tested under most extreme conditions.
Logged

W1BR

  • Member
  • Posts: 4422
Re: 813 parasitic oscillation
« Reply #3 on: July 29, 2021, 07:20:30 AM »

Us old farts would use a neon bulb to detect RF and guess at the frequency based on the color of the neon.  Today we're blessed with cheap spectrum analyzers.
Logged

KI6R

  • Member
  • Posts: 10
Re: 813 parasitic oscillation
« Reply #4 on: July 29, 2021, 10:54:43 AM »

   It would be good practice to identify the frequency of any amp instability.  A grid dip meter makes a good detector of low level RF outputs.  If you bring the dip meter coil near the amp output circuitry while staying safely clear of any high voltages, you will see a sharp peak on the dip meter as you move the frequency dial past the spurious emission.  The dip meter is handy because most will cover all frequencies from low HF to high VHF, so if there is an oscillation the meter will indicate it.  A wideband receiver can also be used.   Solder a 1 or 2" diameter wire loop to the end of a piece of RG-58 to make a pick-up loop antenna.  The loop is soldered between the center conductor and shield.  The loop can be held close to the oscillating circuit and used as the input to a wideband receiver, spectrum analyzer or even a frequency counter.
   Although your amp is stable when terminated in 50 Ohm loads it would be good to test with other loads that represent higher and lower than 50 Ohms, say 2:1 or 3:1 VSWRs.  I suggest adding the parasitic suppressors.  This will help eliminate tendencies to oscillate.  Also, if you ever replace the tubes with different or higher gain tubes suppressors may be needed.
Logged

K6AER

  • Member
  • Posts: 7159
Re: 813 parasitic oscillation
« Reply #5 on: July 29, 2021, 11:04:19 AM »

Terminate the input with 50 ohm load and put a watt meter and the correct 50 ohm load on the output. If the amplifier is truly going into oscillation it will show on the watt meter.

Out of curiosity why do you not want to put parasitic suppressors in the amplifier? The suppressor cutoff shooed be no more than twice the highest frequency of operation.

As you mentioned if the amplifier is always looking at a 50 ohm load this is not an issue.
« Last Edit: July 29, 2021, 11:20:28 AM by K6AER »
Logged

KM1H

  • Member
  • Posts: 11155
Re: 813 parasitic oscillation
« Reply #6 on: July 29, 2021, 11:43:22 AM »

Quote
It would be good practice to identify the frequency of any amp instability.  A grid dip meter makes a good detector of low level RF outputs.  If you bring the dip meter coil near the amp output circuitry while staying safely clear of any high voltages, you will see a sharp peak on the dip meter as you move the frequency dial past the spurious emission.  The dip meter is handy because most will cover all frequencies from low HF to high VHF, so if there is an oscillation the meter will indicate it.

+1 for the GDO and the better ones also include a diode mode (aka wavemeter) for sniffing RF. Mine is an antique Measurements 59 and it hasnt failed me yet for any purpose.

Carl
Logged

WD4HXG

  • Member
  • Posts: 396
Re: 813 parasitic oscillation
« Reply #7 on: July 29, 2021, 12:40:13 PM »

I personally would not be comfortable with assuming any load on the output would insure stability as there is a lot of combinations of reactance and resistance between "0 + or - j inf" and "50 +j0" that may set up conditions for instability.

Regards

Chuck
Logged

K0IZ

  • Member
  • Posts: 1176
Re: 813 parasitic oscillation
« Reply #8 on: July 29, 2021, 06:45:03 PM »

Thanks for comments.  Loaded amp into dummy load, no problems.  Connected to antenna (1.06:1 SWR), made a couple of contacts.  About 10db gain.  At 375 Ma plate, grid current is 80 Ma and peaks with output. 

Pi-net values were to be about 43pf (total with tubes), 3.17 uh coil, and 265 pf loading, Q of 11.5.  Into my Bird 50 ohm load max output however occurs with 350pf (max) loading, and drops off with less capacitance.  Another mystery.

One of my favorite pieces of test equipment is my Millen GDO (tube type).  I have two and find them invaluable.
Logged

W1RKW

  • Member
  • Posts: 219
Re: 813 parasitic oscillation
« Reply #9 on: July 30, 2021, 01:51:23 PM »

not 100% sure what the maximum grid current rating is but 80mA seems awfully high.
Logged

K0IZ

  • Member
  • Posts: 1176
Re: 813 parasitic oscillation
« Reply #10 on: July 30, 2021, 02:05:43 PM »

Most articles I have seen about the 813 GG show 100MA grid current (for 400 to 500 Ma plate or so).
Logged

W1RKW

  • Member
  • Posts: 219
Re: 813 parasitic oscillation
« Reply #11 on: July 30, 2021, 04:02:37 PM »

 I'm running an AM plate modulated 813 (x2) by a pair and run them at 40mA to 50mA so a different beast. it has been said that the 813 is under rated. Glad you got her going. 
Logged

KM1H

  • Member
  • Posts: 11155
Re: 813 parasitic oscillation
« Reply #12 on: July 30, 2021, 05:23:49 PM »

Ive built many amps with 813's and yes the original graphite plate ones are well underrated. Many tests show them with a real world ICAS CW/SSB 200-250W PD before color shows.

With 4 tubes in GG and ~3100V they did 1500W for years here without complaining on CW/SSB; a small fan kept anode seals cool. That amp is still in use down in MA. Others have run them at 3500V without problems and since they were well used for years in BCB AM service that is understandable. The later sheet metal anodes are as bad as 811/811A's :P

I suspect they were initially built to a military spec as their release date coincides to several military TX just prior to WW2. This was common practice; check out the 803 specs against the physical bottle and anode!

Unlike some other tubes they liked more drive/grid current on AM/CW but I suspect SSB linearity suffered.

Carl
Logged
Pages: [1]   Go Up