The AM Forum

Hello!  I am wrapping up my 813 grounded grid "junk box" amplifier, which is my first HF amplifier build.  The amp is described in a February 1969 issue of 73 Magazine: https://archive.org/stream/73-magazine-1969-02/02_February_1969#page/n11/mode/1up 

I had a 20VAC 10 amp transformer lying around and thought I could simply use the center tap to obtain 10VAC (leaving the other leg unused); however, I just realized that a center tap to ground is required!

Since I am using the existing center tap to obtain 10VAC from 20VAC, can I simply create an artificial center tap with a choke (2.5mH) or 10 watt resistors on either side (i.e., the center tap and used secondary that are presently wired to the 813 filaments)?

I hope this makes sense!

Thank you,

Chris, N4JOY

As long as you can supply 10Volts at 10 Amps why would you need a center tap, unless you are monitoring plate current?

Phil

Hi Chris,

Congrats on the 813 project.

Well, unfortunately the RF chokes would burn up due to the fil being only 60 Hz. It's almost like putting a short circuit across the winding, depending on the RF choke's DC resistance. The DC resistance will also cause some bias problems as mentioned later.

Now, using two low value resistors for the center tap is also an idea - and it is sometimes done as a hack. But there IS a price to pay...

This is a linear amplifier and any resistance in the cathode circuit will cause bias fluctuation as the plate current changes and will cause distortion.  The question is, how low in resistance can you make the resistors and not have them draw too much power across the fil winding - but at the same time not create excessive bias variation?  (voltage drop across resistors)

Let's say 5 ohm resistors X2  (across the fil winding) = 10 ohms.  P = V squared / R    =    10 X 10 VAC   / 10 ohms =  10 watts of resistor heat dissipation.  We can live with that.  

Now the bias fluctuation: Let's say with 500 ma peak plate currrent at 5 ohms (2 resistors in parallel) =  E = I X R  =  0.5 X 5 = 2.5 volts of unwanted bias change.  

You need to ask yourself how much distortion/splatter will 2.5V of bias variation cause on a pair of 813s in GG linear service?  The bias (if any is used) should be rock solid stable. I would not do it and err on the side of cleanliness. Find a 10V transformer with a regular "near zero ohm"  center tap for the cleanest signal approach.

Hope this helps.

T

The transconductance of the 813 is 3,7 mMhos, so the cathode impedance will be approx 200 Ohms I don't think that a 5 ohms resistance in series will be noticed. I should go for two resistors of 10 - 25 Ohms

Thank you for the replies.   I have already mounted and wired the transformer (and want to keep this a junk box project), so I was looking for a solution to pulling it.  I guess I could purchase a new 10VAC 10 amp transformer--perhaps Hammond?  I don't have the same luck as others on this forum finding used transformers.  

I found a Hammond 167S10 on eBay... 10 C.T. at 10 amps.    Might be easier to just pull the 20 C.T. transformer than messing with resistor values and hoping for the best.

I really appreciate the group's input.  I will post pics of the amp in the near future!

Chris, N4JOY

If your transformer primary is 220, feed it 110.

Havr a variac you can out in the fil xformer?  Dial it in as you want.


Phil, it's a grounded grid linear amp.  How do you provide cathode return without a center tap?  Their is no external cathode pin in an 813 is their? Without the center tap you can't take the tube out of cut off.



Another idea is doing what Henry does.  Run the filaments in series and ground the common connection in TX.  My Tokyo Hi Power amps (both of them) are this way, using 10v xformers.  On tx, the common point between the two tubes is grounded, providing for cathode return.

This, also, has inherent problems.  Like if one tube is a current hog, a single tube opens up the amp is dead, etc.

But, run whatcha brung!

--Shane
KD6VXI

Thank you again for the responses.  I have one more question!

I secured a 10VCT 10 amp on ePay for a decent price... I am building a voltage doubler circuit for my power supply and the only plate transformer I have available is a smallish 1000VCT at 250ma.  Here is the schematic of the power supply from an old 73 magazine article: https://archive.org/stream/73-magazine-1965-05/05_May_1965#page/n39/mode/1up

I will only be able to drive the amp with about 18-20 watts from my Lettine 240.  I am okay with reduced output and would be happy with a 100 watt carrier for now.  Will this transformer and drive be adequate?

Thank you,

Chris, N4JOY

Hi Chris,

Good on the plans.

I look at it this way... if there is any way to improve or prevent distortion or IMD deterioration, I will do it.  There is nothing worse than someone coming on freq and complaining we are splattering down the band.  It can happen when local conditions are hot and others are trying to pull in a DX statuion. We can be S9 +60 over at their QTH while the DX is S5.

BTW, a pre-distortion "Pure Signal" SDR  system will make most any bad amplifier clean. The following applies only to old school exciters/amplifiers without feedback...


So here's the things that can cause poor IMD numbers and improvements we can make:

1) A poorly regulated HV supply.  A voltage doubler has poorer regulation than a conventional choke input supply, so we lose a few DB of IMD cleanliness there.

2) Resistors in the fil CT will add SOME distortion, maybe not much, but it is preventable. (Thanks for the cathode info, Nico)

3) Improper loading (loading too light) on the finals making more power out, but more distortion. (use LESS C2 capacitor mesh)


4) Neglecting to run 2-tone tests to find ALL  the IMD sweet spots that can improve IMD figures greatly.


5) Running tubes that are less than perfect, low emission, etc., to cause poorer IMD

6) Making sure all DC voltages, grid/ screen/ plate are as solid/stable as possible.

7) Arcing of any kind in components or anywhere in the rig - or in antenna tuner or on antenna itself -  causing CW-arc-gap  simulations... :-)

.8.) Are you using a tuned cathode input circuit? This can add a few DB of better IMD figures.

9) And probably the MOST important and overlooked.. how clean is your driver/exciter? Is it properly loaded/ matched into the 813s?  The 813's final composite output IMD will always be WORSE than the driver's IMD. We want a driver that is 5-10 dB cleaner 3rd IMD than the finals. That is not easy and takes a lot of effort. This is a story in itself.

There's more, but you get the picture.  By watching every detail, big and small, you can end up with an amplifier that is cleaner than the average one on the air. If we neglect some of these things, we will end up with an amplifier where we must be careful where we operate.

You have a great GG design and are on the right path so far... I would get the parts for a conventional choke input 2800 VDC power supply, check the items listed above and rest assured you have done everything you can to make the cleanest system possible.

T

Excellent advice and information, K1JJ.  I will certainly do my best to follow you recommendations--thank you!

Here are some photos of the work in progress.  Yes, the chassis is an old cake pan!  I am sure there are a million ways it could be improved, but this is my first amp building venture; and I wanted to follow the article as close as possible.  I am sure I will be making improvements down the road.  

A few more pics... Again, I thought it would be fun to follow the author's article verbatim and see how well it works.  Probably several things I should have done differently, but we'll see if it works!

I know what it is and have either worked on or designed more amps than you will ever know, but the OP didn't say what the requirement was for the center tap, which is why I asked the question if it was being used for monitoring plate current or bias or what?

I think a CT transformer is probably the best approach if the op follows the original schematic.


Phil


PS So Shane, P**s off!

Chris,

The HV transformer that you have, 1000VCT at 250ma is smallish, as you say. You'll maybe get 30% plate efficiency at carrier and 60% efficiency at full power. If so, then the DC input power would be 333W to get 100W carrier output and 666W input to get 400W on peaks (at 100% modulation). I think your transformer is going to run hot - if it can make the 666W input power at all. You'll find out when you tune for maximum output. Leave room for a bigger transformer.

With around 2000 volts on the plates you can expect about 9 to 10dB of gain if you use a tuned input. 10-15W of drive should do it with a tuned input. Leave room for a tuned input. Better yet, start with one.

Keep an eye on those doorknob caps that you have in tank circuit. They sometimes run hot with a lot of RF current through them. If your tank tuning drifts, suspect those caps. They may heat-up and change value, but you're probably ok at this power level.

Best of luck and have fun. 813's are the best!

Don

Remember Chris…. in the voltage doubler  the power remains the same so your 250ma now becomes 125ma at twice the voltage. And, actually, a doubler will give you a bit more (1.4x) voltage so keep the power rating of the transformer in mind.

I converted a Johnson Thunderbolt amp to grounded grid years ago ... It has the same problem you described that is a fil xfmr with no c.t. .... no other real problem with the transformer so I did the following :

                                 |------------------------------------- to fil xfmr                          1n4007 for the diodes
                                 |---------->|------|
                 ----           |                        |                                                          10 ohm 2 W for the resistors
                 |   |--------|---------res------|
                 |   |                                   |------ c.t 
                 |   |--------|---------res------|
                 ----           |                       |
  cold end of fil            |---------->|------|
      choke                   |-------------------------------------  to fil xfmr

with a bit of other work (vacuum cap plate tune , 3600 V b+, fil choke) the amp would make 1600W with 4-400 pulls and good ota reports

The way you presented your statement / question was in no way like that.  It was questioning the need, not why he wanted one.

To wit: "As long as you can supply 10Volts at 10 Amps why would you need a center tap, unless you are monitoring plate current?"

He needs one for the amp to operate.  Don't get your panties in a wad because you made a simple mistake.  Everyone does.  Lol.

As to the rest of your post, whatever.  I'll not drop to your level nor will I debase myself by replying to anymore of it.



And to another post on the voltage doubler.....  I believe a doubler is 2.8 time the rms AC voltage.  1.4 would be a standard full wave bridge, no?

--Shane
KD6VXI

With plate voltages much above 2000 there may be a need for some form of bias to keep the static plate dissipation down to a reasonable level.

Don

Here is an idea, peel back some of the insulation, count the turns, scrap a little varnish add a center tap lead, doesnt have to be a large size lead wire. Walaaa, problem solved. I have done this on several occasions.
Regards,
Gary

I've done the same thing.

Good idea, Gary.

I just thought of another simple solution...  Just connect the transfomer up as-is using the C.T. already there.  This is 20VAC  CT... Put a small Variac on the primary and adjust the voltage to 10VAC under load.  I always use a Variac on my filament transformers anyway to give a precise fil voltage and I can turn the filament voltage up SLOWLY each time, thus longer life, so WTF.  Why didn't I think of that before?   duh.

(Just looked back and see Shane suggested it yesterday)    double duh.

T

Variac or filaments in series would get him going with what he's got.  The common connection between the two tubes is the 'artificial' center tap as well, if he doesn't have a Variac.

Remember if you use a Variac to ensure that your filament voltage is at 'rated voltage' under key, not at idle.

Care and Feeding by Eimac goes into more details.

--Shane
KD6VXI

Being home brew, a meter in the filament circuit would be the thing to do. You can keep the voltage at where it should be with the variac and at 1/2 voltage on the transformer, you can be assured, it would never burn up. Would look good to.

Shane
 Wawasee does that on the JB-2000 six meter as well as on the 10 80. They add a third winding to the filament choke. The center tap is tied to the mid point of the filaments thru the third winding . I think it is supposed to help balance things out. It works I don't ever see one tube coming up before the other. I am sure you've seen one or two of these.

73 John N8QPC

A Variac can be hard on filaments during on turn on, due to the very low impedance offering high current into a very low resistance filament (until it heats up).  Many BC transmitters use rheostats for that function as they essentially offer a 'soft-start' for the tube.

Jus' sayin.

73DG

Also current limited current transformers using a leak core like a microwave transformer. The Henry generators use these

Here is an alternative approach to a filament choke.  Use a parallel pair of #14 or #12 formvar.  Slip about 25 73 mix beads over each member of the pair.  That will get you around  800 to 1.2K of resistance at the frequencies you will use.

A 12.6 CT transformer, or two 6.3 CT with secondaries in series will do nicely, if combined with a small variac.

I published the choke idea (From W2DU) in a 1987 73 issue.  October I think, but not sure.

Hola Dennis!  How's things out west?

Question: If we feed a filament transformer with the Variac and turn the Variac up slowly each time we turn it on, wouldn't this solve the cold surge problem?  Or are you referring to something else?


BTW, speaking of series filaments, a horror story:  I blew out three 6146 filaments, one at a time, not knowing any better.  I once had four 6146s in parallel - and their 6.3V filaments were in series using a 25.2 fil transformer.   I was testing a group that were hot and took out one or two tubes at a time to replace them with cold tubes. The cold tube filaments resistance was low and dropped most of their the fil voltage across the other tubes - and popped one of the other filaments right away. I saw a bright flash in the hot tubes and then nothing. I didn't realize why this was happening until I popped three in a row...  yikes!  Moral of the story - always heat up series tube filaments together.   Parallel filaments do not have this problem.

The other solution would be to use a Variac on the transformer to SLOWLY bring the hot and cold series tubes up to the same temp.  Still, risky biscuits.

T

I'm puzzled by this, if filament V is adjusted under load, wouldn't filament voltage climb when unloaded?  Why not set V at unloaded condition instead?

An Antek AS-1206 has dual 6v secondaries. Wire those in series, run a few turns 'backwards' through the core (or unwind a few turns, if you dare) and you'll have 10vct. With more or fewer turns, you can 'tune' the voltage. On Epay now for $29 shipped.

Or, your 20volt@10amp filament transformer would work nicely for a pair of Ruskie GK-71's... or three. Just sayin'.
http://www.mif.pg.gda.pl/homepages/frank/sheets/018/g/GK71.pdf (http://www.mif.pg.gda.pl/homepages/frank/sheets/018/g/GK71.pdf)

Another idea. Find a 24 volt secondary transformer and series up the 24 volt winding with the primary of the transformer you have. Measure the output voltage on the transformer you started with. See if it comes close. If not, try  another transformer to buck the primary. Dont use the secondary of output transformers as there are not enough turns. As for the variac, I never had any problems with soft or hard starts. Just be sure the variac is grounded to the amp. If its not, expect all kinds of weird stuff. Be sure the variac is rated at least twice the rated current that you need. Three times is even better.
Regards,
Gary

There would be no real sin in putting a big resistor in the primary of a 20VCT transformer to make it 10VCT@ the desired current. waste of 50W per tube though. hams have done weirder things and will continue to do so in the future.

Yes it will rise between key up and key down.

However, you want to set your tube operating conditions up so that in a directly heated (filament) tube you are at the point of necessary emissions.  Running the tube colder will cause distortion and you won't be able to attain cathode current levels needed.  Finding the point of necessary heat to attain the emissions necessary keeps the tube cooler under operation as well.

If the tube is operated at 'best conditions', key down, and the filament climbs a half volt on key up, who cares (not Eimac!)....  If the tube is cut off, you aren't extracting electrons from the cathode, so you don't need to worry about it.  Note, I'm not saying make sure a 5 volt filament is at 5 volts key down and climbs to 8 volts key up!

If you run it higher voltage levels during key DOWN, then yes, you can see more power out.....  Lots in some cases.  But you will run out of emissions earlier.

Hope that made some sense 🤔

--Shane
KD6VXI

Thanks Shane.  I makes sense.

I just want to thank EVERYONE for their input.  I have much to digest and will continue work!  I hope to have some updated photos and descriptions posted shortly.  MANY thanks to all!

Chris, N4JOY