Call Search
     

New to Ham Radio?
My Profile

Community
Articles
Forums
News
Reviews
Friends Remembered
Strays
Survey Question

Operating
Contesting
DX Cluster Spots
Propagation

Resources
Calendar
Classifieds
Ham Exams
Ham Links
List Archives
News Articles
Product Reviews
QSL Managers

Site Info
eHam Help (FAQ)
Support the site
The eHam Team
Advertising Info
Vision Statement
About eHam.net

   Home   Help Search  
Pages: [1]   Go Down
  Print  
Author Topic: Correct methodology for biasing class AB1 modulator?  (Read 4991 times)
KC9KEP
Member

Posts: 208


WWW

Ignore
« on: January 14, 2013, 05:44:47 AM »

Could someone please help to understand more about this topic or kindly direct me to an appropriate resource?  (The ARRL handbooks don’t go into much detail about this.)

I’ve recently constructed an AM modulator using 807 tubes.  The original design calls for a -22.5V “hearing aid battery” to be used to bias the tubes.  I had used (2) 9VDC and (3) 1.5VDC batteries in series to generate a bias of about -22.5 VDC.

However, my power transformer provides a higher plate and screen voltage than what the original design had called for.  Also the original design suggests that the bias be set so that the cathodes indicate about 50 mA of current draw at idle (The cathode current is monitored by measuring across a 50 Ohm cathode resistor.)

Although my output waveform looks very nice on the scope, the 807’s are running hot and I was seeing some plate redness.  And, measuring the voltage across the cathode resistor, I had calculated about 125 mA of current draw.
So I reasoned that if the plate and screen voltages are higher than specified in the article, the bias should probably be more negative as well.

So, I tried (4) 9VDC batteries to yield a bias voltage closer to -36VDC in an attempt to bring the cathode current down to the specified 50 mA.

At -36VDC bias, the cathode dropped perfectly to 50 mA and I thought all was good.  But .. looking at the modulated wave on the scope, the wave shape doesn’t look as sinusoidal as it did with the -22VDC bias .. the rise and fall of the waveform looks more “straight” than curved.

So, then I reasoned that although I’ve “corrected” the cathode current to 50 mA, this may not be the appropriate bias for my higher-than-called-for plate and screen voltages.

Now, I am wondering how this balancing act should be correctly performed.

Should I simply add an adjustable bias supply that allows me to adjust the bias somewhere between “too hot” and “poor linearity” while monitoring output on the scope?  (Maybe trapezoid test?)   Or is there a specific method or tables indicating how to setup the operating bias for a push pull class AB1 amplifier?  Maybe the idle current needs to be higher than 50 mA when running the modulator at higher than specified plate/screen voltages?

(Note, to check the performance of my modulator, I have been connecting it to my 6146B transmitter; modulator transformer in series with the 6146B plate supply, and to 6146B screen via dropping resistor as indicated in the construction article.  I have been using 1 KHz audio test signal.  I monitor transmitter output into a 50 Ohm “Cantenna” load via a current monitor, a small torroidal transformer.)

Thanks all!
--KC9KEP
Logged
KC2ZFA
Member

Posts: 15




Ignore
« Reply #1 on: January 14, 2013, 07:53:24 AM »

why not put a variac on the power supply to correct the voltages, verify the article's design, and then experiment with higher voltages and a variable bias supply...don't forget that you shouldn't exceed the
maximum plate dissipation at rest.
Logged
N2EY
Member

Posts: 3860




Ignore
« Reply #2 on: January 14, 2013, 01:48:04 PM »

It's not about just the bias. It's about the operating conditions of the tubes.

The plate current of screen-grid tubes is much more dependent on the screen voltage than the plate voltage. Since your power supply gives too-high screen voltage, the operating conditions are all wrong.

What you need to do is:

1) Look up the recommended operating conditions for the 807 in Class AB1 at the plate voltage your power supply gives.
2) Redesign the screen circuit to provide the recommended screen voltage.
3) Redesign the bias circuit to provide the recommended grid bias.

Only then will it work right.

IIRC, screen voltage on an 807 should never exceed 300 volts. The design shown uses only 230 volts. What voltages is your supply giving?

73 de Jim, N2EY

« Last Edit: January 14, 2013, 01:51:02 PM by N2EY » Logged
N2EY
Member

Posts: 3860




Ignore
« Reply #3 on: January 14, 2013, 01:57:04 PM »

Found the other thread. Nice job - but the use of silicon rectifiers will give you much higher B+.

This isn't a big problem because 807s will work up to 750-800 volts on the plates. What's needed is to get the screen voltage down where it belongs.

btw, you can put the unused filament windings to work as a bias supply. Some diodes and caps in a voltage-multiplier will do the trick. AB1 draws no grid current so it's easy.

73 de Jim, N2EY
Logged
N2EY
Member

Posts: 3860




Ignore
« Reply #4 on: January 14, 2013, 06:32:46 PM »

However, my power transformer provides a higher plate and screen voltage than what the original design had called for.  Also the original design suggests that the bias be set so that the cathodes indicate about 50 mA of current draw at idle (The cathode current is monitored by measuring across a 50 Ohm cathode resistor.)

This is a detailed follow-up to my previous posts.

First, the usual method of doing this sort of thing is to look up the recommended operating values for the tubes in use and design around them. However, there isn't much direct info on 807s in AB1.

So here are some general rules:

1) The no-signal plate and screen power inputs should not exceed the maximum values for the tube type. For the a pair of 807s, you have a total of 60 watts plate dissipation and 7 watts screen dissipation. In practice it's a very good idea not to push the bottles that hard, particularly in an AM modulator, where the duty cycle is much higher than in a CW or SSB transmitter.

So let's say 50 watts plate dissipation and 5 watts screen dissipation - absolute maximum

2) The plate current will rise dramatically with screen voltage.

3) The screen voltage should be reasonably well regulated

4) Once you deviate from the original design, you probably have some design and modification work to do.
 
Although my output waveform looks very nice on the scope, the 807’s are running hot and I was seeing some plate redness.  And, measuring the voltage across the cathode resistor, I had calculated about 125 mA of current draw.

Which is way beyond tube ratings! If you have, say, 700 plate volts, 125 mA plate current is 109 watts dissipation!

So I reasoned that if the plate and screen voltages are higher than specified in the article, the bias should probably be more negative as well.

So, I tried (4) 9VDC batteries to yield a bias voltage closer to -36VDC in an attempt to bring the cathode current down to the specified 50 mA.

At -36VDC bias, the cathode dropped perfectly to 50 mA and I thought all was good.  But .. looking at the modulated wave on the scope, the wave shape doesn’t look as sinusoidal as it did with the -22VDC bias .. the rise and fall of the waveform looks more “straight” than curved.

That's because there are probably other things that need changing. - screen voltage, plate impedance, etc.

So, then I reasoned that although I’ve “corrected” the cathode current to 50 mA, this may not be the appropriate bias for my higher-than-called-for plate and screen voltages.

Now, I am wondering how this balancing act should be correctly performed.

Should I simply add an adjustable bias supply that allows me to adjust the bias somewhere between “too hot” and “poor linearity” while monitoring output on the scope?  (Maybe trapezoid test?)   Or is there a specific method or tables indicating how to setup the operating bias for a push pull class AB1 amplifier?

There is a specific method. Fortunately it's pretty easy.

  Maybe the idle current needs to be higher than 50 mA when running the modulator at higher than specified plate/screen voltages?

NO!! Just the opposite!

Remember Rule 1 - you must not exceed the rated plate and screen dissipation! If anything, you want LESS plate current at higher plate voltage!

(Note, to check the performance of my modulator, I have been connecting it to my 6146B transmitter; modulator transformer in series with the 6146B plate supply, and to 6146B screen via dropping resistor as indicated in the construction article.  I have been using 1 KHz audio test signal.  I monitor transmitter output into a 50 Ohm “Cantenna” load )

The problem there is that the modulated stage may not be a good matched load, and may not be linear. At this stage of the game, you'd be better off making up a dummy load of equivalent resistance using power resistors. For example, if the 6146B is running 600 volts at 120 mA, (72 watts in) the load resistance should be 600/.12 = 5000 ohms. The modulator needs to produce half the power of the stage input - 36 watts - so you need about a 40 watt resistor. Better yet, a combination of smaller resistors.

The scope patterns are not the best indicator of linearity until you have done the basic stuff.

Here's how I'd proceed:

1) Find out the actual B+ voltages your supply is putting out. The design values were 600 volts and 230 volts, but obviously you're getting more. You need to know how much more.

2) If the low B+ is 300 volts or less, and the high B+ is 750 volts or less, you're in luck - the original circuit is OK. All you need to do is adjust the bias and the plate load impedance.

3) In the original design, they ran the tubes at 600 volts and 50 mA. That's 30 watts static plate-and-screen dissipation - a good number to remember. Note that the 50 ohm resistor measures combined plate and screen current, not just plate current.

The best info I could find for 807s in AB1 is the following (values are for TWO tubes):

Condition 1: 600 plate volts, 300 screen volts, 34 volts grid bias, 36 mA resting plate current, 0.6 mA resting screen current, 140 mA maximum plate current, 16 mA maximum screen current, 56 watts maximum output

Condition 2: 750 plate volts, 300 screen volts, 35 volts grid bias, 30 mA resting plate current, 0.6 mA resting screen current, 140 mA maximum plate current, 16 mA maximum screen current, 72 watts maximum output

Other operating conditions may be interpolated. Note how these values involve higher voltages and lower currents.

The original design called for a plate-to-plate load impedance of about 12,500 ohms. This is for 600 volts on the plates, 250 on the screens, 50 mA idling cathode current and 100 mA full signal. With higher voltage and lower current, the load impedance will be higher. Selection of taps on the mod transformer is the way to obtain a match between the modulator and the modulated stage. The transformer data sheet tells how.

Hope this helps

73 de Jim, N2EY 

Logged
KC9KEP
Member

Posts: 208


WWW

Ignore
« Reply #5 on: January 15, 2013, 05:55:16 AM »

Jim,

This is all great information .. thank you for taking the time to put this together!

I'll to get back to the project later this week.  I had neglected to mention what
voltages that I had initially measured (with the initial -22.5VDC bias), but here they are:

B+ Supply (Modulator off) = 810VDC
B+ Supply (Modulator on) = 732VDC

Low V Supply (Modulator off) = 316VDC
Low V Supply (Modulator on) = 285VDC

(The original schematic says B+ should be 600VDC and L.V supply 230VDC.)

Voltage across 50 Ohm Cathode resistor (Modulator on): 6VDC
Cathode Current (Modulator on) = 6VDC/50 Ohms = 120mA

Of course, the cathode current had dropped to 50 mA when I increased the bias
to -36VDC.

But, thanks again, I will return to this project soon.  (Maybe it would be easier
to replace the existing power transformer with one that has lower B+ output?)

73

KC9KEP
Logged
G3RZP
Member

Posts: 4393




Ignore
« Reply #6 on: January 15, 2013, 05:58:19 AM »

If it's capacitor input power supply, just removing the reservoir capacitor will drop the voltage, although you will need a bleeder.
Logged
N2EY
Member

Posts: 3860




Ignore
« Reply #7 on: January 15, 2013, 06:39:17 AM »

If it's capacitor input power supply, just removing the reservoir capacitor will drop the voltage, although you will need a bleeder.

In the original schematic, both sections are choke-input. They used a 5V4 and two 6X5 rectifiers in the original "economy" circuit, but KC9KEP used silicon rectifiers.

A very similar modulator appears in some of the 1950s ARRL Handbooks.

73 de Jim, N2EY
Logged
N2EY
Member

Posts: 3860




Ignore
« Reply #8 on: January 15, 2013, 06:50:06 AM »

Here's what I would do:

1) The plate voltages are high but usable - no changes to the B+ needed.

2) The unused 5 volt rectifier filament winding would be put series-aiding with the 6.3 volt heater winding, giving 11.3 volts or so. This would drive a voltage tripler (three diodes, three capacitors, whole thing on a subassembly) to develop about 50 volts of bias. (Be sure to use a voltage tripler circuit that has one side of both the AC input and DC output grounded)

3) A voltage divider would be used to reduce the bias supply output to 35 volts or so. This is OK because there is no grid current in class AB1 operation. Bias would be set to give about 45-50 mA total cathode current with no signal. I'd aim for 45 mA. (Would also be absolutely sure the 50 ohm resistor really is 50 ohms).

4) A dummy load equivalent to the modulated-stage load would be built for testing, using power resistors. At audio it doesn't matter if they're wirewound. Power rating at least 60% of the transmitter modulated-stage DC input.

5) Using the transformer data, the appropriate taps would be selected to give at least 12,500 ohms plate-to-plate, and hopefully a bit more. Testing would then be done into the dummy load. 1000 Hz sine wave is a good test. Care must be taken not to overdrive anything.

73 de Jim, N2EY
 
Logged
WA7PRC
Member

Posts: 160


WWW

Ignore
« Reply #9 on: January 16, 2013, 11:11:26 PM »

Here's what I would do:

2) The unused 5 volt rectifier filament winding would be put series-aiding with the 6.3 volt heater winding, giving 11.3 volts or so. This would drive a voltage tripler (three diodes, three capacitors, whole thing on a subassembly) to develop about 50 volts of bias. (Be sure to use a voltage tripler circuit that has one side of both the AC input and DC output grounded)

If you follow the 5V + 6.3V + tripler with a LM317HVT (VIN-VOUT = 60V max), you could have adjustable and regulated bias!  Reference the LM117HV/LM317HV datasheet.  Digi-Key & Mouser both have the LM317HVT (TO-220 package) in stock for a coupla bucks or so.  Smiley

Even the regular LM317T could be used but, you would have less headroom.

vy 73,
Bryan WA7PRC
Logged
N2EY
Member

Posts: 3860




Ignore
« Reply #10 on: January 17, 2013, 02:30:06 AM »

If you follow the 5V + 6.3V + tripler with a LM317HVT (VIN-VOUT = 60V max), you could have adjustable and regulated bias!  

Yes, but you don't really need regulated bias in Class AB1. There's no grid current, so the load never varies. Variations in line voltage also cause variations in the plate and screen voltages, so unregulated may actually be better.

73 de Jim, N2EY
Logged
N3QE
Member

Posts: 2094




Ignore
« Reply #11 on: January 17, 2013, 09:34:49 AM »

As a different design point... The audio guys in the late 50's and 60's used beam tubes with high resultant linearity, with the plate voltage much lower (400v?) And the screens hooked to taps on the primary on the plat transformer with a dropping resistor. I think they call this "ultralinear".

This is a good ways away from the 50's ham or spec sheet advice of regulated screen supplies and higher than spec sheet plate voltages. While the amplifiers are similar, the power supply iron ends for one is a poor match for the other.
Logged
N2EY
Member

Posts: 3860




Ignore
« Reply #12 on: January 17, 2013, 09:56:27 AM »

As a different design point... The audio guys in the late 50's and 60's used beam tubes with high resultant linearity, with the plate voltage much lower (400v?) And the screens hooked to taps on the primary on the plat transformer with a dropping resistor. I think they call this "ultralinear".

Pretty close. The Ultra-Linear concept used tubes such as the KT-88, 6550, or EL34/6CA7, and a screen-tapped output transformer, or more-conventional tubes with a tertiary screen winding so that the screens weren't at the same B+ level.

This is a good ways away from the 50's ham or spec sheet advice of regulated screen supplies and higher than spec sheet plate voltages.

The modulator in question doesn't use higher than spec sheet plate voltage.

The big difference is that a modulator for a ham rig is basically a voice device, not a hi-fi thing.

73 de Jim, N2EY
Logged
K1ZJH
Member

Posts: 901




Ignore
« Reply #13 on: January 25, 2013, 08:33:31 AM »

If you follow the 5V + 6.3V + tripler with a LM317HVT (VIN-VOUT = 60V max), you could have adjustable and regulated bias!  

Yes, but you don't really need regulated bias in Class AB1. There's no grid current, so the load never varies. Variations in line voltage also cause variations in the plate and screen voltages, so unregulated may actually be better.

73 de Jim, N2EY

Another possibility, since the tubes are AB1 and we have extra plate and screen voltage, is to use a back bias method to generate the grid bias for the audio output tubes.

Pete
Logged
N2EY
Member

Posts: 3860




Ignore
« Reply #14 on: January 25, 2013, 10:40:37 AM »

Another possibility, since the tubes are AB1 and we have extra plate and screen voltage, is to use a back bias method to generate the grid bias for the audio output tubes.

If by "back bias", you mean having the negative supply not grounded directly, that can work. However, it has some problems which make the separate bias supply a better idea:

1) The bias will vary with total load current, and since this is AB1, the load current varies.
2) There is a considerable amount of rewiring needed to make the change, while the voltage-tripler can be made up as a module and installed without much disruption of the existing stuff.

73 de Jim, N2EY
Logged
Pages: [1]   Go Up
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.11 | SMF © 2006-2009, Simple Machines LLC Valid XHTML 1.0! Valid CSS!