The AM Forum

I am in difficulty to decide what is the most reliable bias sircuit.
What do you think between A or B and what are the positives/negatives of each one?

Stefano

Hello Stefano,

What a coincidence. I'm currently in a same situation where I have to choose between the two.

I opted for "A". I can't explain why, but my feeling told me (And maybe I'm very wrong) that it would be nicer to have a short ground path through the Gridleak resistor, rather than through the negative voltage supply. My negative voltage supply is on a different deck.
My worry is the (grid)amp meter. The idea is to place it between the resistor and ground.

So... you're not alone in asking the question... ;)

73, Martin PA4WM

Using circuit A you probably will want to put the meter between the RFC and the grid leak resistor.

Fred

Sircuit A is also my favorite but I would like to know if the result of the diode between the negative supply and the grid leak resistor is to keep the safety bias away during transmitting periods so all the bias comes from the grid leak resistor and to allow the safety bias to reach the tube during receiving.
In this case is it critical the amount of the safety bias let's say 1/3 of the total or anything but less than the grid leak bias under full drive or anything?

Stefano

With either one Id use a 1.5 to 3K non inductive resistor across the RFC to minimize any chance of LF parasitics if its a high value or in band ones if its a low value.
Remember that an amplifier is just a failed oscillator.

Carl

LOL !

" Remember that an amplifier is just a failed oscillator. "

Carl is so cruel.

The amplifier is not a failed oscillator, it is just a challenged oscillator, in which reasonable adjustments are made. H.S. Black would understand.

klc

Hello Stefano,

I have actually used circuit "A" in a working amp. The only difference is that I used a full wave center tap rectifier circuit. I'm not sure what is the purpose of the other diode next to the grid bias resistor, but the circuit works and works well.. I suspect it would also work without that other diode, but I left it in place.

To figure out what value of grid leak resistor to use, it is a simple matter of E=IR. The safety bias value should be somewhere near the tube cutoff value, which can be determined from the tube data sheet provided by the manufacturer, or determined experimentally. I think that -100VDC would be a good place to start for most of the commonly used tubes with 1500-3000VDC on the plate. As I understand it, when there is no grid drive, all the bias is provided by the bias supply, and when grid drive is present, all the bias is provided by the grid leak resistor.

I built two different working amps, one with a Penta 254W as the final, and one with an RCA 8000 final. Both had about
-100VDC safety bias, and then of course had the appropriate value of resistor to provide grid leak bias. I did put a DC voltmeter between the grid leak resistor and the DC ammeter, and it gave me readings that were exactly what I wanted. With no grid drive, -100VDC and zero grid current. With grid drive at proper amount to give recommended grid current, and resistor chosen by E=IR, grid voltage was right on. BTW, I have used both 1N4007 and 1N5408 diodes sucessfully.

I see no reason that circuit "B" would not also work well, but I think that the grid leak resistor value would have to be chosen experimentally. Hope this helps.

Ron W8ACR

PS if you do a search on "LF parasitic oscillation" on this website, there is a thread by that title started by myself and there is a hand drawn schematic that I posted that partially shows my bias circuit. this might help.

 Circuit B has a problem. Here the bleeder resistor shown, but with no value given, is the only path to ground.

The diode in the power supply will get back biased whenever the grid current exceeds the bleeder current. This 

makes the grid leak bias contribution more complex to calculate. Therefore circuit A is preferred.

Another option is just a grid leak resistor, and a overload relay. I still like circuit A.

Jim
WD5JKO

Reasonably low distortion practice for bias in class C amps that are to be plate and screen modulated is to have the bias voltage split approx. equally between a regulated supply and a grid leak resistor in series.

For example an 813 "wanting" -150 volts bias on its grid might have -75 volts regulated and, say, a 10 k to 13 k grid resistor plus an RF choke of 2 to 5 mh for Rf isolation.

  This is true, and this method is practical for RF tubes like the 6146 where the grid current is small. Keep in mind that the regulated bias supply you mention either needs to be a "shunt" regulator, or a simple supply that has bleeder current higher than the grid current. A simple supply as shown in 'B' will work provided the point I made is met. A series regulator won't work at all since it can only 'source' current when it must 'sink' current. A shunt zener or vr tube properly sized, and biased, are examples that are applicable.

  Circuit 'A' has an advantage since the bias supply does not have to 'sink' current, and once RF excitation is made, the circuit is 100% grid leak. Like Ron says, this works, but is just one way to do this among many.

Jim
WD5JKO

I have not found this to be the case and have used grid leak as the entire bias source of plate modulated class C amplifiers with good results.   One of the best class C finals that I made used a pair of 813's with only grid leak bias and the screen supplied by a dropping resistor from the modulated B+.  In this case a relay coil was in series with the grid resistor who's contacts  prevented the application of plate voltage until proper grid current reached.  This arrangement seemed to work better than the "clamp tube" that was also tried.  This particular final modulated better then any other tetrode that I have personally used.  I also had similar results with a pair of V70D triodes.  In my experience grid leak bias alone works fine as long as there is some other provision to protect the tube in the absence of grid drive while plate voltage is applied.

I think "B" needs a shunt VR to avoid the PS floating up due to grid leak current developing a voltage across C.

I vote for "A."

Yes, in amateur practice, having only a grid leak resistor works well since the average grid current increases rapidly as the plate voltage is reduced to zero, pretty much self regulating.

But now I quote, "Electronic Designers Handbook," Landee, Davis & Albrecht, McGraw-Hill, 1963?:
"To obtain the best modulation linearity, a portion of the bias for the grid is obtained from a fixed bias supply and part is obtained from  a grid-leak resistor.  The values of (each) which will provide the best linearity over the complete modulation cycle are detrminded closely by finding the values of fixed and grid leak bias which locate the operating line BB' at the crest of the modulation cycle.... "  etc.

And then references further a diagram of operating lines showing load lines both unmodulated, modulated at peak with optimum ratio of fixed to grid-leak bias. 
The half and half voltage split of the bias' is a first approximation.

Quoting further, "By using only grid-leak bias, the rf output voltage will increase more rapidly with increasing plate oltage than desired for linear modulation.  The linearity can be optimized in practice by adjusting the relative values of fixed and grid leak bias while observing  the modulation Lissajous figure as shown in Fig. 5.6b and 5.6c. " 
The Lissajous figure shown is exactly what we call a scope trap figure showing the modulation factor "M" = (B-A)/(B+A)  where "A" is the left smaller voltage of the truncated prism and "B" is the rightmost larger voltage of truncated prism somewhat reduced from the constant slope line. 

 Well, you all know the source of trap voltages on your modulaton scopes so I won't go into more detail other than to say if correctable distortion showed up on "ancient"  CRT's of '33, '43, '53 and '63, then it must be noticable audible, heh, heh. 

Reasonably best practice in that book referred to commercial and amateur practice. ..   The same book goes into great detail about such things as screen bypass capacitors, noting that the value should only be enough to bypass RF without undue phase shift between the plate and screen during modulation, etc.  All kinds of good stuff.  Wonder if I can adequately scan it for the archives, copyright stuff and all?  One of several problems is that sections refer to each other in interleaved fashion.

 

That sounds like a good idea, as the grid draws current it would kick the relay on and as soon as you lost grid drive, it would kick the relay off. Now I imagine the HV relay was between the power supply and the final, right? If it was on the primary to the plate iron, then it could be possible to still have some HV draining off the filter caps even after the relay was switched off, which would kinda defeat the purpose of the whole setup since there would still be HV on the plates after you lost grid drive.

The grid current relay contacts were in the lead to the plate supply relay coil and also in series with the contacts on a small relay that keyed the exciter.  The HV was switched in both these transmitters.  It provided a slight delay of the HV to allow grid drive to come up and the antenna relay to switch.  Yes, there may be an instant between grid drive going off and the B+ bleeding down.  However I used the 813 transmitter for years without issue so this arrangement was satisfactory in my case. 

It seems that many plate modulated class C finals have some compression of the positive peaks, especially when modulating in excess of 100%.  So if  "By using only grid-leak bias, the rf output voltage will increase more rapidly with increasing plate voltage", then this may effectively compensate for peak compression to some degree.

In the case of my old 813 rig, I speculated that the high degree of audio applied to the screen grid helped it modulate so well.  But maybe the grid bias arrangement was part of the reason too.

regaarding the original question, I like circuit "A" too.

another way with grid leak bias but no fixed bias to keep the final from self-destructing if you loose drive would be to use a small amount of self-biasing to keep the final at cutoff when there's no grid drive. with a directly heated tube like an 813, a rheostat placed between the filament transformer center tap and ground should work by adjusting the rheostat unti you get the amount of bias you want.

I dont care for either and I really do not like the 50/50 idea. Its best to use only the exciter to bias the final for the least distortion.

But as OUK pointed out, If you loose Drive, The finals will see full current and you could damage them.  There are several ways to deal with this.  One is a simple plate current Trip like the T368 has.  If the final runs over that set plate current, The HV is Tripped. 

The other is to use a circiut like some BC rigs that will provide a constant bias to the final, When Drive is detected above and beyond that fixed level,  the fixed bias drops away.  I think gates used this method?  I cant remember which one. 

Hams use Grid leak 50/50 because its simple and cheap.  If a HAM can save a dollar its a major victory.  Performance is sacrificial.

C

Hey Clark,

If I understand your post correctly, you advocate for 100% grid leak bias. Two things - i believe that in circuit "A", if the grid leak bias is significantly more negative than the bias supply, essentially all the bias voltage comes from the grid leak. Secondly, I like your idea of a plate current trip. What does that circuit look like? Could you direct me to a schematic of such a circuit?

Thanks, Ron

If the plate overload circuit is the way I think it is, it uses an relay with the coil in series with the B+ to the plate, and the contacts to the power supply. If too much current passes through the coil, it opens the contacts, which shuts down the plate supply.
Shelby

I can take a picture of the t3 schem later. Its pretty simple.  It uses a resetable breaker.   

OK. It is agreed that Class C bias needs to be split between protective bias, which will take the PA tube to cutoff with no drive, then the rest of the needed bias will be provided through a resistor.

Now, bear in mind- That the tube's grid current with drive will need to be dissipated *somewhere*. For a Class C pentode or tetrode that doesn't need too many watts of drive, I suppose a bleeder on the DC bias supply will be OK. Like an 813, 4-125 to 4-1000.

But if you are dealing with a Class C triode, that needs a lot of drive, then you need some sort of bias regulator circuit that can dissipate the grid current.

Here's what I used in my 304-TL triode amp. Grid current is disposed of through the 6BQ5 regulator tube, which holds the fixed bias voltage constant. Circuit works like a champ. Would also be good for use with a pentode or tetrode amp.

You can use almost *any* triode or pentode regulator tube capable of handling the PA's rated grid current. Anything from a 2A3 to a 6AQ5. First tube circuit I have ever used that could accommodate almost any tube.

In this circuit, R7 is in line to provide proper Class C operating bias. Or it is shorted out to be a linear amplifier for SSB (which I have never used). You don't need to use a hollow-state bias supply and the 6X4 could be easily replaced with a couple of solid state diodes.

Bill

Circuit A does just that.  All the grid bias comes from the grid leak resistor.  If the grid leak bias drops below the fixed bias the connecting diode maintains at least the fixed bias on the final.  Once the grid leak bias rises above the fixed bias the diode is biased off.  Fairly simple circuit.

Fred

Fred. The Fixed bias is there all the time though right?   The Grid leak just goes over that fixed level.  Unless I am reading this wrong.

C

Seems like if you are going to use a triode as an RF tube, it would make sense to use an 805, 811, or 572B or some other zero bias tube.

Then you can use a grid leak to get class C, but if the driver fails you end up back as the class B idle current.

No regulated bias supply needed.

Dave

The part you have to look at is the diode linking the bias supply to the grid circuit.  The fixed bias supply is on all the time but when the grid leak bias is more negative than the fixed bias the diode is biased off, so it's like the bias supply is not there.

Sometimes it's harder to see things when you're dealing with negative voltages.  Another thing you have to consider, with that circuit, is that the grid leak resistor acts as a load resistor on the bias supply when there is no drive.  So that 5K grid leak resistor pulls about 50 ma from the bias supply (at -250V).  If there was some drive (but not enough) the 5k resistor would load the bias supply less.  Once there is enough drive to produce -250V there would be no load on the bias supply from the 5K grid leak resistor.  Once the drive produces more than -250V the diode is reversed biased and it's like the bias supply is not there.

So, the bias supply has to able to supply some current and maintain the -250 volts.  Another thing is that, with no drive, the tube doesn't draw any current from the bias supply.  The grid draws current on the positive peaks of the drive signal.  As the grid becomes more negative the grid is drawing current for a shorter duration of the (pos 1/2) sine wave.  As you can see the class C final is really a pulse generator.  It's the tank circuit that produces the full sine wave output signal.

It was really hot here in NJ today.  My brain is somewhat melted, but I think everything above is correct.  I welcome any corrections from others on the forum.

Fred

  The fixed bias supply is on all the time but when the grid leak bias is more negative than the fixed bias the diode is biased off, so it's like the bias supply is not there.
[/quote]

Does that mean the gridleak resistor has to be calculated for the total amount of negative voltage needed, instead of just the part to add to the fixed supply?

Martin

Ups... something went wrong with the quote... sri

Does that mean the gridleak resistor has to be calculated for the total amount of negative voltage needed, instead of just the part to add to the fixed supply?

Martin

Ups... something went wrong with the quote... sri
[/quote]

Martin,

You're exactly correct.  The grid leak resistor is calculated for the total negative grid voltage.  Don't forget to figure the power dissipated in the grid resistor.  You can use any combination of resistors to get the power rating and resistance.

Remember I'm referring to circuit A in the original post.

Fred

Yes, exactly.
In the A case the grid leak resistor must be calculated suitable to provide the total bias. I changed a bit the sircuit in order to get the possibility of B or AB class.
I will never use my C class as a linear amp but I have heard that keeping a little idle anode current (100ma in my case) in expense of some power during receiving periods keeps my reactor peaceful when I cut off the driver. I use to keep the anode high voltage on during receiving so handling the reactor's energy calm is a special task.

Stefano

Its always a good idea to bias any Class amp to about 2X cutoff when in receive/standby.

First it eliminates any diode noise in the receiver and secondly it offers much better protection from a parasitic or neutralization type oscillation.

Carl

Thanks for the clarification. I see how this works now with the diode. That might be the system that I was thinking of in the gates transmitter.  If the fixed supply is not there during the time the exciter is on, That sounds like a great system.


C