Print Page - homebrew amp schematic

Hello all,

Attached is a schematic of a homebrew amp that I have designed and built. I have borrowed heavily from K9ACT's design of his 8000 tube amp. (thanks Jack) Before I apply power, I wanted to solicit comments on the design. This amp will be driven by a Viking I. It will be a single band 40 meter amp and will be modulated with a pair of zero bias 811A's at 1000V plate voltage. It will run Eb 1800V @ 180 mA and Eg -240 @ 40mA. I would particularly like to hear comments on the design of the grid bias circuit and the Plate tank circuit. I am concerned about parasitics and proper neutralization. Hopefully I can get the schematic upright this time. I'll try to post a picture of the amp later.

Thanks, Ron W8ACR

Ron,

I look forward to seeing the pictures. A project like yours can be so fulfilling when it all works as intended.

I might look further at that bias. As shown the bias supply will ratchet up with grid current. If you applied excess drive, say 100ma upon keyup, the bias would go to -600v (E=0.1A X 6K). That might hurt that 450v rated filter capacitor. I would consider isolating the bias supply with a diode (cathode to the right). Also that 6K resistor could be a 10K or 12K WW rheostat. Having adjust-ability here will allow you vary the drive power needed for a given amount of grid current.

Good Luck!
Jim
WD5JKO

Hi

Looking at the schematic,it's good,and hopefully works well.
But to prevent VHF and UHF parasitic oscillations why don't You used a parasitic choke between the plate of the tube and the 2.5mH Plate RFC.
the parasitic choke consist of non inductive resistor between 51 to 100 Ohm with a coil in parallel with it,
the coil can be 6 t0 8 turn of no 14 wire ,wound around the resistor body( but now it is told to used the coil in parallel with the resistor,not wound around it.}.

and use a low value resistor 10 to 50 ohm in series with the tube input near the tube socket.

Gito.n

Ron

The modulation resistance of the tube will be 1800 volts / 0.18 amps = 10,000 ohms.

But, the impedance of the 5000pF (0.0050uF) rf bypass capacitor on the B+ line is 10,000 ohms at a frequency of 3200 Hz. This means that the modulated B+ will be 3dB down at a frequency of 6400 Hz as a result of that capacitor [assuming the modulator has a source impedance that is also 10,000 ohms].

In addition, the 1000pF plate coupling capacitor adds its capacitance to ground at audio frequencies... via the path to ground through the pi network inductor and the safety choke.

You should consider reducing the sum of the rf bypass capacitance and the plate coupling capacitance to no more than 3000pF.

At 7.3MHz, a 1000pF capacitor has an impedance of 21 ohms. Based on that, I would suggest that you consider using 1000pF for the plate coupling capacitor, and 2000pF for the rf bypass capacitor on the B+ line.

Stu

Quote from: w5hro on April 28, 2010, 12:59:43 PM

Way too much extra junk on the bias supply line.

Why not just convert it to a grounded center tap supply with the normal choke input circuit with the appropriate bleeders, etc.?

It seems to me the capacitor input filter he has now should be OK. Wouldn't hurt to throw in a second 100uF filter cap (making it 200 uF total).

But, why have an RF choke and bypass cap on both sides of the grid current meter?

Seems like having the choke and bypass on the grid side would be sufficient.

Ron

Your pi network does not appear to have the correct component values for 7.3MHz operation.

If you use the 7uH inductor shown in your schematic, you will be operating (when the tube is loaded properly) at a Q of 18. This will lead to high losses in the tank circuit.

With the tube operating in Class C, if you redesign the pi network for a (typical) Q of 10, you will need the following nominal component values

Inductor: 11uH (The 7uH value you have now is too low)

Tuning capacitor: 44pF (The current choice of a 150pF variable capacitor is okay)

Loading capacitor: 23pF (The current choice of a 1200pF value capacitor may have too large a minimum value... taking into account the output capacitance of the tube and the capacitance to ground of the various interconnections.)

However: because of your choice of 1800 volts of plate voltage and 0.18 amps of plate current (at carrier)... if you redesign the pi network for a Q of 10, the lowest antenna load resistance that you will be able to properly load into is: 50 ohms. You will not be able to properly load the amplifier into an antenna load that is less than 50 ohms, even if you set the loading capacitor to its lowest value.

If you redesign the pi network for a Q to 12, then the nominal component values become:

Inductor: 9.9uH

Tuning capacitor: 50pF

Loading capacitor: 248pF

This would allow you to match (load into) an antenna load that is somewhat lower than 50 ohms (i.e. 35 ohms) by reducing the loading capacitance. With the loading capacitor set to 248 pF you will be able to match a 50 ohm antenna load. If you increase the loading capacitance toward 1200pF, you will be able to match an antenna load impedance of significantly more than 50 ohms.

I suggest that you redesign the pi network for a Q of 12.

Stu

Hi

As far as I know the min bias for a RF tube is to protect ,to minimize the standing current of the tube when there's no drive,
that means We "don't need" this min bias supply ,when there's the rated grid current and the right Grid resistor when there's RF drive to this tube.They called it Automatic bias.

So my conclusion the min bias is needed when there's no drive to protect the tube.
When there's a RF drive which it's " right" we "don,t" need this "protective" bias anymore.

And Contrary ,when there's RF drive.The min bias power supply is not discharged but Charged by the rectified RF voltage by the tube.

For instance in this circuit it used a protective bias is - 110 V,
Than when it was driven ,there's a 40ma grid current flowing in the tube and the min bias soars to (Rg)6000 X (grid current).04= min 240V.(operating bias)
So the min bias Supply is not discharged but Charged with the rectified high frequency (RF) voltage drive

So the question is .... Do we need a well filtered and stiff min bias power supply for A class C RF amplifier? (not for Audio amplifier}

Gito.N

Hi

What I'm trying to wrote,That A class C Transmitter need's no min bias supply,as long there's a RF drive that's needed and a Grid resistor of the correct Value.
Like this 8000 RF tube--- with 40ma grid current and 6000 ohm grid resistor We got a min 240 on the grid 1 that's the correct value of the operating Tube's bias.

So what the purpose of the min bias supply?
As far as I know it is used only to protect the Tube when the Drive is gone, limiting it standing current/current drawn with no RF drive to the tube.

So since it's only a protective bias ,any min bias supply can be used as long it can supply the needed current that flows in the Grid resistor.when there's no drive and the voltage needed ,to limit the standing current to a save value.

Gito.n

Hi

Of course a well regulated power supply is the best,But what I'm trying to explain is the importance of it. Do we need it?

If We looks at the old commercial build transmitter for the amateur ,a lot of it use Automatic bias (no min bias supply)
,looking at BC transmitter,they don't used regulated min bias.but a min bias heavily loaded with a R bleeder.

I'm not against it ,So if we can used a regulated power supply it's good,
but as we know the purpose of the min bias ,I don't see a fault at the min bias of the 8000 schematic.

Gito.N

I am having trouble following this thread. Some questions keep coming to mind.

Why do we need a stiff regulated bias supply to protect a class C RF amplifier from a loss of excitation? All we need is enough bias to keep the tube below rated Pd in the absence of excitation.

The circuit as shown with 100ma grid drive (overdrive burst) will put 600V across the 450v bias capacitor. Won't this hurt the capacitor? Remember 100ma * 6000 ohms = 600 volts..

Isolating the protective bias with a diode does work, and then when the RF amp is running, the bias is all grid leak so long as the grid leak bias is more negative than the fixed bias, which it is at 40ma grid drive.

Why not eliminate the bias supply all together, and instead use a latching overload relay?

With a plate modulated class C RF amplifier with grid leak bias, does the grid leak resistor need to by un-bypassed for the modulating frequencies? I've never seen a large electrolytic across a grid leak resistor before.

sorry for asking too many questions. :-\

Jim
WD5JKO

Thanks to everyone for your posts. I have made some changes to the circuit, especially in the bias supply. Brian, I can't figure out exactly what you are describing in regard to the use of a zener diode. Perhaps you could draw a schematic? :) Does anyone think that a 2.5mH choke in the grid circuit will cause an LF parasitic problem?

I have attached a new schematic.

Thanks again, Ron

Hi

The ARRL handbook stated .....when RF choke are used in both grid and plate circuits of a triode amplifier,the split condensator tank combine with RF chokes to form a low-frequency parasitic circuit,unless the ampliifier circuit is aranged to prevent it.

The Grid RF choke is substituted by a Resistor at least 100 Ohm.(from the ARRL handbook.

Here's a schematic of Gates Radio Company BC1T a 1 Kilowatt Am broadcast transmitter that uses automatic bias/no bias power supply in the driver(two 6BG6G)
and final ( two 833).

Gito.n

Mike,

My power supply is a simple full wave bridge rectifier using a string of four 1N4007 diodes per side each diode in parallel with a 470K resistor and a .01uF ceramic disk cap. The filter is a choke input using a Stancor C1403 swinger and a C1413 smoother. Filter cap#1 is 3uF@3000V and #2 is 12uF@2000V. Plate transformer has no center tap and secondary is rated 1800VAC at 270 mA. I am measuring 2000VAC at the secondary terminals and 1800VDC at the 100K bleeder so I'm sure it will be somewhat less under load. I have attached the pertinent part of the PS schematic. It does not show the 470K resistors or the .01uF caps.

Ron

Ron

I think there is another problem (easy to fix) with your current design (refer to the updated schematic you posted).

The problem has to do with "self biasing" of the RF output tube.

To achieve self biasing, the RF component of the grid current... that flows when the RF input signal goes through a positive peak, once each RF cycle... must flow through a capacitor. That way, the capacitor charges up on each input RF cycle... and the grid leak resistor removes this charge in the form of a continuous, average "grid current".

In your design, the "self bias" voltage builds up on the two 140 pf variable capacitors. That is, every time a pulse of current flows (on each RF cycle) into the grid of the tube from the top of the RF transformer, it returns to the bottom of the RF transformer through the lower (in position on the schematic) of the two 140pF capacitors. Thus, that capacitor (and the upper one as well) builds up the negative self biasing voltage

However, in your current design, the grid of the tube is coupled, from the perspective of DC, directly to bias supply.... and its associated large capacitors.

The time constant for changing the voltage across these capacitors is 32uF x 6000 ohms = 0.19 seconds.

Thus, it will take a long time for the self-bias to reach steady state.

To fix this problem, you need to put the 6000 ohm resistor in series... between the grid and the fixed bias supply (not in parallel with the supply; as in the current design)

Stu

Quote from: AB2EZ on May 03, 2010, 09:14:04 AM

To fix this problem, you need to put the 6000 ohm resistor in series... between the grid and the fixed bias supply (not in parallel with the supply; as in the current design)

Stu, with all due respect, I agree with your analysis except for the solution. Putting the grid leak resistor in series with the bias supply won't work because there will be no return path to ground. The bias will shoot way up with excitation. A possible solution is to make the bias supply a shunt regulated supply with a 10W 110 volt zener diode with a current limiting resistor (from the power supply). The zener could be three 36v 5 watt jobs in series to get close.

Jim
WD5JKO

Jim

You are correct... since the grid current is flowing in the direction that would charge the capacitors, my suggested solution would not work.

The fact that the grid current actually pumps power into the bias supply, rather than drawing power from the bias supply (pointed out by Gito, you, and others) is one of those perennial sources of confusion.

I agree that one way to provide a return path to ground would be to place a Zener diode in series with a resistor across the output of the bias supply; and to connect the grid leak resistor to the junction of the Zener diode and the series resistor.

Perhaps a simpler way would be to leave the grid leak resistor as it is in the current version of the schematic (going directly to ground), and to add a 1N4007 in series with the bias supply and the top of the resistor (instead of directly connected). That way, the grid bias will be at least as negative as the bias supply (i.e. the grid will not be able to be less negative than the fixed grid bias supply, because the diode would be forward biased)... but when grid drive is applied, the grid voltage would be able to move more negative than the fixed bias supply without having to charge up the capacitors in the bias supply.

Stu

Hi

This is a schematic of my simple min bias supply.(A)

Look at B ,in this circuit like WB6IYH wrote wont work ,because when one of the VR tube has a slight difference on triggering ,the one that has a higher voltage when triggered . Triggered first and the voltage drops a across the 3000 ohm resistor at the junction of R3 and the 100 ohms drops to a low level,that makes the voltage to low to trigger the other VR tube.
At least at this circuit because A VR tube needs at least 40 t0 50 VDC higher
than it's operating condition ,like VR105 it needs 150 VDC to trigger it.

At (C) the VR tubes are directly drop from the power supply ,which each dropping resistor,
So the voltage drop across each dropping resistor won't affect the operating condition of each VR tubes.

Gito.N

Hi all,

Thanks for all the comments. I think I'll just try the bias circuit as I have drawn it and see how it works. I am attaching two schematics, one is my hand drawn rendering of the bias supply circuit from the original Globe King 500, the other is from K9ACT's schematic of his 8000 tube amp. Both bias supply circuits are essentially the same as mine. If I have problems, I'll add the diode as suggested by Jim and Stu. If it still doesn't work, I'll ask for more help.

One more thing. Will an LF parasitic act the same as a VHF parasitic? Can I expect runaway plate current with both types of parasitics? If an LF parasitic is present, can it be cured by changing the value of the bypass cap which is adjacent to the RF choke? Or should I just remove the RF choke and replace it with a resistor?

Thanks again, Ron W8ACR

hi.

From QST.

Regulator circuit for loads that may vary as much as 70 ma.the two VR tubes are indentical.R1 and R2 are chosen so that 40 ma is drawn by each under minimum load conditions.
That means R1 and R2 is in kilo ohms value. not 100 ohm and directly taken from the B+.

But I see that You have succeed in building you min bias supply.
When using 100 ohm equalizing resistor ,it means to equalize the current flowing trough the VR tubes not the triggering voltage of those tubes.And the same VR tubes because the manufacturing variations,can have different triggering voltage.

From GE application ......operating of glow tubes in parallel is not recommended

Why still used Choke filters,in the old days they used choke filter because it's hard to find high value Capacitors ,in now days we can find 220uf 450VDC easily,so without a choke ,the DC voltage is well filtered
And it is used to feed a regulator tube that ads to the filtering of the supply? so I believed it well filtered min bias.

Gito.N

Brian,

Using grid leak bias on a class C amplifier is ideal. The correct operating point (conduction angle) is then functions of the grid leak resistance, and the grid current. This is a time tested no compromise solution.

Where the debate comes in, as it seems, is to protect the Class C tube in the absence of excitation. Here we can use an overload relay. I think the BC-610 does this. We can also provide a means of protective bias to keep the tube below ratings without excitation. Playing with the ratio of fixed bias to grid leak bias can compromise class C operation.

I just don't see any need to regulate anything here with zener diodes, or VR tubes. Doing so complicates the circuit, and also may compromise the class C tube performance.

My vote is the overload relay with grid leak bias. This approach will also give you the benefit of protecting the tube from high off-resonance plate current as well as plate current that goes high for any other reason such as a parasitic.

If the overload relay is undesirable for some reason, then grid leak bias with diode isolated fixed bias has already been proposed. Here when the drive is at the proper level, the bias is 100% grid leak -> ideal. Then with low drive, or NO drive we have fixed bias to protect the tube.

Now, if this class C tube was also going to run class B as a linear amplifier, then regulating the grid bias (Low Z source impedance) is something to look at. Here we want load regulation, but NOT voltage regulation.

Just one mans opinion.

Jim
WD5JKO

Quote from: w5hro on May 15, 2010, 04:23:14 PM

By the way, how would using a single 0B2 tube or a zener diode and then one dropping resistor complicate the grid circuit?

Good question. Let me take a stab at it. The big Eimac tube wants -240v @ 40ma grid current (1st post this thread). So if we use a bias supply along with fixed bias resistor, as Stu pointed out, the supply must be in series with the grid leak resistor. Doing this we lose our path to ground because the direction of current is opposite what the diodes allow (in power supply). We fix this with a zener diode, or a VR tube as a shunt regulator where we connect this device between ground, and the junction of the grid leak and the bias supply. The shunt device must have a series resistor back to the bias supply to limit current. For a VR tube we need to maintain at least 5 ma minimum. Now when we fire up the grid drive at 40ma, this is additive to the VR current from the bias supply, so maybe we have 40+5=45ma when an OB2 is good for 30ma. Also the grid leak resistor needs to provide 240-105=135 volts to get -240. Using a zener instead of a VR has to do with the thermal issues of a zener, i.e. a 5W zener means 5W when the leads are attached to an infinite heatsink at 25 Celsius. In real life, a 5W zener is really like a 1w zener. We can deal with this by adding many LV zeners in series...

Maybe I'm full of it, but this scheme seems to complicate things.. ;)

Jim
WD5JKO

Brian,

The neat thing about ham radio, and hobbies, is that we don't have to design in a way to pacify the bean counters. Making a 'one-of' prototype can use outrageous strategies because we want to experiment. So there is no need for us to always follow the KISS principles. (KISS = keep it simple stupid)

As an example, I took a Sherwood 30+30W stereo amp with FM, and applied the 'regulate everything' approach. The results were a fantastic boost in unit performance even though the concepts and approach might be ridiculous.

So here is an example of one of my anti-KISS FOS designs:

http://pages.prodigy.net/jcandela/Sherwood_S8000/HV_REG/S8000_Mods2.htm

Been going 7 years now....almost daily usage.

Jim
WD5JKO

Quote from: w5hro on May 16, 2010, 01:42:25 PM

But where are the vacuum tubes :-\

Well, the tubes still do all the audio amplification as before. I just modernized the power supply with SS since it would all fit on a 2"X2" PCB using an old CPU heatsink with fan. Doing the same thing with tubes would result in a chassis the size and weight of a Hammerlund SP-600... ???

So as a tube jock, do you think that is cheating? If so, do the two diodes in the bias supply (original post) need to be replaced with something like a 6X4? :-*

Jim
WD5JKO

The AM Forum

THE AM BULLETIN BOARD => Technical Forum => Topic started by: W8ACR on April 28, 2010, 01:35:31 AM