Heathkit SB220 intermodulation

[W4AMV]

Good day forum.

I am investigating this for a colleague as well as for my own information.

The Heathkit SB220 applies negative feedback to the 3-500Z tubes via a combination of RFC and decoupling C to the 3-grid pins. See attached per manual. Is this effective as compared to no degeneration and tying all the grids hard to DC ground?

Thanks, Alan

[Steve - K4HX]

IIRC, the IMD spec for the SB-220 was around -30 dB PEP. That is no better than many other amps using a pair of 3-500Zs and not better than the numbers on the Eimac spec sheet.

[W4AMV]

Hi Steve.

Thanks for the feedback (no pun intended).

So if I understand your reply, the circuit technique used by the Heath engineers did not accomplish much to improve IMD and consequently, direct grounding of all grids of the 3-500Z tube is perfectly acceptable.

Alan

[W2VW]

Direct grounding of the grids increases gain slightly. Lots of creative writing on this subject on the WWW.

Cathode drive still has feedback built in to the design.

Grounding those grids directly should be coupled with a glitch resistor in the
B+ line as it removes a safety feature.

[K1JJ]

Hi Alan,

As Dave said, the inherent GG NFB is already there and works very well. (Feedthru power in the cathode)

Yes, the common engineering opinion is to directly ground the grid with copper strap to the chassis. It will create a more stable RF path on the higher bands. I doubt those components in the grid create a worthwhile RF negative feedback path.  

I usually add in the adjustable bias in the form of zeners or diodes in the cathode lead at the fil xfmr center tap.

Over many years using 3-500Zs and other tubes in linear service, I have always gone directly to ground with the grids. My chain of class A / AB1 system 8877s easily does -45dB 3rd order IMD when tested using grids directly tied to ground.

Another thing NOT to do is put a resistor or choke with large resistance in series with the cathode of an AB1 / B  linear amplifier expecting beneficial NFB. It will get worse.   This technique works only for pure class A and is common in old tube audio amps.

Remember that when you drop linear amplifier power down by 1/2, the 3rd order IMD improves by 9dB. This is incredible and a reason to have plenty of headroom and run the amplifier far from peak compression level. I have heard 3-500Z amplifiers on the air that are run balls to the walls and splatter like heck. We need to back them down and load them heavily (less  C2 messed) in order to be clean. Just because they say "Henry" or "Alpha" or other commercial brands does not make them run clean when pushed too hard.

Some amps have their own internal RF NFB circuits which is a plus. Still, someday when pre-distortion software becomes main stream, the common open loop 3-500Z type ham amps will sound like spark gaps in comparison...

T

[W4AMV]

Thanks fellows for all the information!

Alan

[W4AMV]

Thanks fellows for all the information!

Alan

Direct grounding of the grids increases gain slightly. Lots of creative writing on this subject on the WWW.

Cathode drive still has feedback built in to the design.

Grounding those grids directly should be coupled with a glitch resistor in the
B+ line as it removes a safety feature.

Oh, I meant to ask... Can you clarify the "glitch resistor" item in your post. Are we trying to limit plate current upon loss of drive? As I would think the cathode network would assist there.

Thanks, 

[K1JJ]

Oh, I meant to ask... Can you clarify the "glitch resistor" item in your post. Are we trying to limit plate current upon loss of drive? As I would think the cathode network would assist there.

Thanks,  

A power resistor of 10 - 25 ohms in series with the DC plate lead will help reduce surge currents in case of a catastrophic short in the plate circuit. Loss of drive will not hurt a GG linear amplifier. Maybe you are thinking of a class C stage without protective bias?

Speaking again of NFB for a GG linear amplifier.... for every one dB of NFB, it takes one dB more of drive power to the amplifier for the same output power. So if we wish to reduce our IMD by -6 dB with 6dB of NFB, it would require an additional X4 of drive power.  This would be a problem for most since the amplifier stage is now down around 10dB of total gain for many popular power grid tubes.

IE, the negative feedback inherent in a standard GG amplifier stage is more than plenty to give a reasonable IMD starting point with 13-16dB of stage gain. (without additional NFB added)

T

[W4AMV]

Ah, got it Tom and Thanks again!

[Tom WA3KLR]

I always wondered if the parallel L-C network grid to ground was for gain compensation - 80 meters versus 10 meters; more gain on 10; net gain constant across all bands?

The parallel L-C network is resonant at 200 kHz.

[DMOD]

I always wondered if the parallel L-C network grid to ground was for gain compensation - 80 meters versus 10 meters; more gain on 10; net gain constant across all bands?

The parallel L-C network is resonant at 200 kHz.

The DCR of the choke used is 23.5 ohms and the more grid current that is drawn, the higher the bias on the grid.

It acts mainly as a fuse and as a grid bias.

[K1JJ]

The DCR of the choke used is 23.5 ohms and the more grid current that is drawn, the higher the bias on the grid.

It acts mainly as a fuse and as a grid bias.

Hmmmm....  for linear operation, wouldn't the DC grid bias swing around based on grid current flow causing poor linearity?   Bias should be stiff at all currents, no?

T

[KD6VXI]

I've argued that point till my keyboard was blue,  Tom....   Along with Jim VE7RF.  People don't wanna look at that side of the coin.

Best thing I've found for the sb220 was to ground all 3 grid pins to chassis,  add more bias voltage,  and hang a BIG cap off the bias source (diodes,  active device,  whatever).

Amp runs happy,  and no bang bath boom events.   Original Eimac,  full power out. 

--Shane
KD6VXI

[KN4SK]

Hello all, I'm the colleague W4AMV was referring to. Two questions for this wise group (and thanks to Alan for posting for me):

1) What voltage rating does the glitch resistor need to have? I've got 25 ohms, 25W. Is the rating on the order of 1A x 25 Ohms kind of range or are we talking hundreds of volts during a fault condition?

2) The reason I starting looking at the grounded grid configuration is because my dad brought his SB220 to me complaining of smoke coming out. Turns out the RFC from grid-GND on one tube is burned out. I'm good with replacing it and the grid caps with bus wire, but do you have any idea why the thing might have burned out in the first place? Pretty much everything on this amp is original from the 60's or 70's other than I replaced the HV cap bank and the chassis mount zener which blew last year.

Thanks for any advice.

Ray

[DMOD]

As for the "glitch" resistor I am sure you will get a number of opinions but as for me a twin 3-500Z Amp does well with a 10 Ohm 5-10W pulse rated or plain old ceramic resistor.

As for the choke burning out, it could have experienced a tube flashover or inter-element short.

As for the bias zener, a 1 Amp SB fused, 7.5V 10W stud zener works well.


Phil - AC0OB

[W2VW]

Hello all, I'm the colleague W4AMV was referring to. Two questions for this wise group (and thanks to Alan for posting for me):

1) What voltage rating does the glitch resistor need to have? I've got 25 ohms, 25W. Is the rating on the order of 1A x 25 Ohms kind of range or are we talking hundreds of volts during a fault condition?

2) The reason I starting looking at the grounded grid configuration is because my dad brought his SB220 to me complaining of smoke coming out. Turns out the RFC from grid-GND on one tube is burned out. I'm good with replacing it and the grid caps with bus wire, but do you have any idea why the thing might have burned out in the first place? Pretty much everything on this amp is original from the 60's or 70's other than I replaced the HV cap bank and the chassis mount zener which blew last year.

Thanks for any advice.

Ray

Most likely cause is a short to one of the grids from somewhere. Cathode shorts are more common.

Second cause gassy bottle.

Third likely cause oscillation.

Note for testing. Some shorts do not show up with the bottle cold.

[w1vtp]

Ken uses a piece of #30 magnet wire as a "glitch" resistor in the HV.  See attached photo from his web site.  This has always been my understanding of a glitch resistor - it has a history of being effective.

http://w2dtc.com/3cx3000f7.htm

Al

[Opcom]

I've used that! It works perfect. There is no resistor to explode with shards, only the wire.

[w1vtp]

I've used that! It works perfect. There is no resistor to explode with shards, only the wire.

Yes, it does work very well, although I think I would enclose the wire in a glass tube affair to limit the "shrapnel" should a glitch happen and the #30 wire explode (yup, that's what it does!!)

Al

[K1JJ]

Yep, the wire works FB.  I use two strands of fine RG-213 braid. And as shown, the posts need to be at least 2" apart to finally quench the arc. For 200 to 500 mS or so there will be a sustained arc until the wire fully vaporizes.

I use BOTH the fine wire and a resistor.  I think we need the resistor too, cuz until the wire breaks there will be full surge current going thru the circuit.

BTW, that looks like California #30 wire....  

T

[w1vtp]

Yep, the wire works FB.  I use two strands of fine RG-213 braid. And as shown, the posts need to be at least 2" apart to finally quench the arc. For 200 mS or so there will be a sustained arc until the wire vaporizes.

I use BOTH the fine wire and a resistor.  I think we need the resistor too, cuz until the wire breaks there will be full surge current going thru the circuit.

BTW, that looks like California #30 wire....  

T

I like your current limit idea with the resistor - might eliminate the exploding wire  

PS: I've used small gage solder in similar applications.  When the magic current value happens the solder just melts and little drops fall.  Not sure it would work for almost 4 KV tho but in lower voltage apps - - -   That was back when I was a snotty nosed kid and didn't have a Radio Shack around to buy fuses

[K1JJ]

Solder sounds like an interesting method, Al.   Though at 4KV I'll bet solder will still arc over into a sustained flame for a moment.

The solution is to use genuine high voltage fuses encased in a vacuum. I don't know if they use a vacuum and have never tried them....  but that would probably make the arc quench almost immediately.

3A @ 10 KV costs about $54 / each. Too expensive for ham use:

http://www.mouser.com/ds/2/87/cooperbussmann_Bus_Ele_DS_6003_HV_Series-350193.pdf


Stop by RatShack and axe for one. Might as well get a trophy blank look before they close...    (Bring a camera)

T

[kg7bz]

I've been wondering about these "High Voltage Microwave Oven" fuses for protecting the transformer in my Johnson 500 from downstream problems. I see them all over, ebay, Amazon, other appliance repair sites, 500mA-1A @5KV. Given that they're only 40mm long, I suspect they'd be best used in the lead from the transformer to the rectifier so the current isn't steady DC so the arc would have a chance to extinguish.

http://www.ebay.com/itm/2pcs-High-Voltage-5KV-5000V-6-x-40mm-Glass-Tube-Fuse-for-Microwave-Oven-/321580505662?

August
KG7BZ

[KN4SK]

Thanks for the replies everyone, I'm good to go. Unfortunately it looks like I have a bad tube. 40 years of very regular use... not too shabby!

[Opcom]

Those fuses have but 20mm/0.8" of space between the metal ends. OK maybe for the HV transformer secondary but are they suitable between the filter bank and the tube/load?