The AM Forum

..HAD A QSO WITH CARL,WA2UJX,ES HE MENTIONED HE WAS USING AN APACHE THAT USES 6146s IN THE MUDULATOR...HE ALSO MENTIONED THESE MODS ARE SIMILAR TO DX100 MODS...HAS ANYONE ELSE DONE THESE MODS?,OR SHOULD I DO THE TYPICAL MODS THAT I SEE LISTED?
 ..TNX..TIM..SK..

Tim,

Last March I was talking about re-doing the audio in my Apache which had already been messed with to the extreme in the modulator section by the former owners. Well it is back on the bench with the PTT working, great RF output and the Power supplies have been "solid-stated". To the modulator we go...

I was trying to figure out what bottles to use and if I wanted to install some more substantial iron. I have been working on a modulator schematic which uses a 12AX7, a 12AU7 phase spliltter and a 6CG7 push-pull driving the finals.
I was considering 6146's, 6146B's, 6DQ5's 6550's, EL-34's and KT-88's and KT-90's. Anyway, I settled on KT-88's because of thier voltage rating and plate dissipation. I mounted the new modulation tranny and put in three sockets which can accept tube shields.

First of all, I wanted to run the low level stage fillaments on DC as suggested by WA1QHQ. This is a hum reduction idea. To light a pair of 12AU7's and a 6CG7 it takes about 1.2 Amps at 6.3VDC. I could have developed this by taking the now unused 5V winding on the  5V4 and making a bridge and putting in a large cap like a 4700 uF but I would still have some ripple. Instead I found a dandy 6.3V DC 2 Amp switcher and RTV'ed it to the side of the end transformer. It runs cool and no ripple. If were to do this again, I would use a 12V switcher and run both the fils and the PTT (a 12V relay instead of a 24V) on that supply. 

Mike WU2D   

Screw up #21 - The Svet KT-88's are too big in the thighs.

Each tube fits individualy but the tube sockets are just too close together. I will have to move one of the sockets over about a 1/16. The EL-34's, 6146s and sweep tubes and SOME 6550's would fit with no issues!   

Mike WU2D

OK - more butchering

The short story is that the KT-88's now fit. The operation required that the octal socket had to be removed (which actually was a good thing since they got cleaned up) and the holes had to be slightly eccentricalized. This was done with a Greenlee punch and some chunks of sacrificial aluminum.

I think that some air will be needed over the mod tubes for obvious reasons not the least of which being that the VFO compartment is right next to them!

The Apache Trail of Tears continues...

Mike WU2D

Mike

Looking at the standard Apache schematic, I would be concerned about the 30k ohm screen resistor. The differential impedance, looking into the screen (i.e., the change in screen voltage associated with a small change in screen current), for a single 6146, varies from less than 5k ohms to more than 15 k ohms as you modulate the screen voltage. For a pair of 6146's in parallel, the differential impedance, looking into the screens, will vary from less than 2.5k ohms to more than 7.5k ohms as you modulate. Thus the voltage divider formed by the screen input impedance and the 30k ohm screen resistor will vary from less than 1:13 on positive modulation peaks to more than 1:5 on negative modulation peaks. For tetrodes, it is best if the screen voltage follows the audio modulation waveform (if you want low distortion) reasonably accurately.

To fix this, you might consider tying the dc screen biasing resistor (e.g., the existing 30k resistor, or whatever value gives you the desired screen DC bias voltage) to the unmodulated B+, and using an AC-coupled, lower impedance voltage divider to deliver the modulation to the screen. For example, a 5k ohm resistor in series with a .5 uF capacitor (or larger) from the screens to ground, and a 15k ohm resistor in series with another .5uF capacitor between the modulated B+ and the screen.

Stu

Great Stu,

I am interested in this of course and when I get to the modulation part, I will be looking at all of your suggestions.

First I have to start building the modulator from scratch.

Thanks,

Mike WU2D

Stu,
I had good luck adding 50 K 10 watt to ground screen load at the tube sockets of the 6146s with a 30 K screen dropping resistor. This dropped my screen voltage down to about 170 volts and reduces the modulator swing on the screens to about 70%. (5/8) This also serves as a bleeder for the HV supply since the big dropping resistor was replaced with the zener regulator for the modulator screens. All I had to get it below the zener voltage was a pair of 500 k loads across the zeners.

Frank

Please clarify:

Since the impedance looking into the screens (i.e., the change in screen current resulting from a change in screen voltage) is much less than 50k ohms... shouldn't the fraction of the plate modulation voltage swing that is on the screens be much less than 50k ohms / 80k ohms?

I believe that in most of these types of designs, the equivalent source impedance of the modulation being directly applied to the screen(s)... in this case, approximately: 50 k ohms in parallel with 32 k ohms* (including the ~2k ohm impedance of the plate modulator as seen on the secondary side of the modulation transformer), which equals ~ 19.5 k ohms... is much larger than the input impedance looking into the screen(s) of the tube(s) being modulated. Thus, the screens are, to some extent, "self modulated" (i.e., electrons flowing from the cathode start piling up on the screen when the plate voltage drops... thus lowering the screen voltage).

*Good old Thevenin's theorem: I knew I would have a use for that some day.

To force the screen voltage to faithfully follow the audio modulating waveform, the screen would have to be driven from a modulating source whose impedance is lower than the impedance of the screen(s) of the tube(s) being modulated (in parallel). In this case, looking at the curves for 6146's, one needs a modulating voltage source impedance of ~ 2500 ohms... much less than the 19,500 ohm modulating source impedance that corresponds to a 50k ohm resistor from screen to ground and a 30k ohm resistor from screen to modulated B+.

Best regards and Happy New Year!

Stu 
 

Thanks
Stu

Yup, You are right Stu. The source Z is still high. I was just trying to drop the screen voltage more. All I did was drag the voltage down. My method would work if the resistor values were lower. Then there would be a lot of load on on the modulator and supply. I wonder if there will be any weird phase shifts with the ac coupled signal and wire wound resistors.
A third transformer winding is the real fix because you want a low source z like 500 ohms.

Frank

Yes... a separate winding with a separate, low impedance, screen voltage bias supply in series (or better yet, a separate transformer, if the plate modulation transformer doesn't have a good low end response) would be best.

On my DX-20, which I have externally plate modulated, I used the approach of AC-coupling the modulation to the screen:

50k ohm resistor from unmodulated B+ to screen (as per original DX-20 design)

Using the "40% ultra-linear" tap on the backward-connected output transformer that I am using to modulate the plate of the DX-20... I connected that tap to the screen with a 1uF capacitor in series with a 13.5k ohm resistor. Since the impedance looking into the screen of the 6DQ6A r.f. output tube is around 15k ohms (less or more, depending upon the screen voltage), the screen modulating voltage is (roughly) 40% x 15k/(15k + 13.5k + the source impedance of the 40% tap of the transformer) ~ 20% of the plate modulating voltage. Meanwhile, the source impedance of the screen modulating voltage is around 15k ohms (13.5k ohms + the source impedance of the 40% tap of the transformer)

The above gives good linearity of modulation... but still not as good as I think I would get with an even lower impedance arrangement for modulating the screen.

I am thinking of adding a 5k ohm resistor in series with another 1 uF capacitor from screen to ground... and replacing the 13.5k ohm resistor (above) with another 5k ohm resistor. This would produce a screen modulating source impedance of ~ 2.5k ohms, while maintaining the 20% ratio of screen modulating voltage to plate modulating voltage.

The concern is probably not the phase shift (the typical boat-anchor modulation transformer has lot's of phase distortion at the low end). The concern I have is that the screen will draw more incremental current on positive excursions of the screen voltage than will be offset by the reduced screen current on negative excursions of the screen voltage. Thus, the AC-coupling capacitors will charge up a bit (shifting the bias point) with modulation applied. However, I think this will be a small effect.

Stu

Imagine designing a color TV with tubes. I'm convinved it took real talent.
So much easier today with Class e and FETs.

Frank

I substituted the 5k ohm / 5k ohm voltage divider into my DX-20: one side of each resistor capacitively coupled to the screen with a separate 1uF capacitor; the other side of one resistor connected to ground and the other side of the other resistor connected to the 40% "ultra linear" tap on the output of the external modulation transformer.

The modulation is very clean, as measured objectively with a sine wave (30Hz - 5kHz with no significant roll off, harmonic products >30dB down, sine wave output of the off-air monitor looks beautiful on the scope), and as determined subjectively by listening to the output of the off air monitor while speaking into the microphone.

Stu

Stu,

I could actually do this on the Apache because the transformer has two tertiary windings, one low Z and one high Z, but we will see when I get there!

Mike WU2D

Well I am in the middle of bringing the Apaches new modulator up and I have not even been able to ge past the MIC amplifier without problems!

First take a look at the schematic of my 12AX7 mic pre-amplifier.

The switch allows two modes - In one postion it is a normal high Z high gain amplifier for a D-104. In the second position the switch removes the mic amp and turns the second stage into a fixed gain grounded grid stage for 600 Ohms input. The switch is located in the 6AL5 tube socket hole right next to the 12AX7. You reach in a flick the switch when you want to change from mic in to line in.

Problems on the next post:

Mike WU2D

The circuit works in both modes and I have all kinds of gain but that MIC GAIN pot is noisy (junk?).

OK here are the real problems that I am having so far. The conditions are as follows: Voltages as shown on schematic and the fils are regulated DC with no ripple and I see no ripple on HVDC either. Only the first tube, the 12AX7 is inserted. Tube shield used.

I had heard that there was an issue with the shielded cable in the Apache which runs all the way across the front of the radio in a big wire harness. The MIC GAIN pot cable also has a long run in the same bundle. I had heard that this arrangement is a hum magnet and the cable is leaky.

I can verify this because I have hum showing which goes away when I kill power. It takes over 30 seconds after killing power before that stage stops amplifying. At this time it is clean as a whistle.

The second problem is that my frequency response goes down to 10 Hz in both modes but my high frequency response 3 dB point is 4 kHz for the High Z configuration and 7 kHz for the grounded Grid. Why the roll off? I checked the input to see if that was being loaded but it is steady with no roll off at the 220 p cap.


Help!

Mike WU2D

Great thread guys........I have an Apache on the way and am watching this topic with great interest.
Bill

Mike

Looking at the schematic (pretty standard for all the rigs of that era), the high frequency roll off at 4kHz is puzzling.

Question: Does the instrument you are using to measure the signal at the output of the preamplifier have a high enough input impedance. The plate resistance of a 12AX7 is something like 62.5k ohms. If your measuring instrument has an input capacitance that is around 500 pF or higher... that would produce the roll off you are seeing

Stu

Stu,

I am using the Leader LMV-181A Millivoltmeter and a scope and both agree. I can try a higher Z approach.

As for the hum, I am going to try running new shielded mini cables for the mic and the pot away from the wire harness on the front panel. I will route them along the inside wall where the controls come though the chassis. 

Next comes the phase splitter stage.

Mike WU2D

Mike

The LMV-181-A meter is specified (assuming the specification sheet I found on the web is correct) at 10M ohms and 50 pF input impedance. So, that, in and of itself, should not be a problem in this measurement application. I don't know what the scope's input impedance is... and I don't know if you have them both in parallel across the measurement point at the same time.

Watch out for the capacitance of any cables used to connect the rig to the measurement instruments. If the cables have (for example) 30pF per foot of capacitance, it would only take 15 feet, total (adding the capacitances of two cables in parallel), to get 450 pF.

Stu

Ok I got some time to hit the bench finally and do some troubleshooting and wiring tonight. I owe Stu a cookie!

I made up a temporary Hi-Z probe out of a 100K and a 1K resistor and attached that to the output cap on the stage. Then I reduced the scale on the AC voltmeter to get back to around a 100 mV RMS output level ( 1 mV after the divider). NO MORE ROLL OFF. It seems that the capacitance of my crappy test cable made from some old audio cable was bad news!  :-*

On the hum front - I replaced the two shielded cables with some nice tightly braided mini-microphone cable and routed them away from other conductors. While I was at it, I replaced the Heath 500K MIKE GAIN pot with a high quality Clarostat. Caution - the panel and chassis require a pot with a thick threaded mounting bushing. This reduced the hum to the point where I think I am seeing only common mode noise from the generator.

Mike WU2D

Tim; Some other speech amp and high level possibilities. Remember Cam-Main  IE      K1DEU/VE8 Cambridge Bay, Victoria Island  Collins Autotune  4-1000 mod  by 4-400's  73  John, We Do not QSL OM

 Electric Radio Article  "Universal speech amplifier/modulator" below,

 http://hamelectronics.com/k1deu/pages/ham/transmitters/am/pages/universal_speech_amplifier.htm

If you want to use 6550's or KT88's but because of the close spacing of the sockets the toobs wont fit, try KT90's. The envelope is the same size as an EL34.
http://www.boiaudioworks.com/index.php?cPath=59 (http://www.boiaudioworks.com/index.php?cPath=59)

    The wide majority of modulation distortion problems occur in the audio driver stage. By using either a push-pull transformer coupled or two direct coupled cathode/source followers overall distortion is essentially eliminated. We rarely can make enough audio in AB1 class and when the modulator tube grids draw the slightest amount of grid (AB2) current the resulting sudden very low impedance reflected to a single stage driver's distortion is easily heard !  The little push-pull 12AU7 is not the least bothered by the wide changing transition between AB1 and AB2 as we speak along !  Open loop corner frequencies are properly established for extremely stable closed loop operation. I.E. No in-band low frequency motorboating or out-band ultrasonics.  Some may find the JFET Phase splitter (option) interesting as there is a space problem in DX-100's, rangers, etc.  73, John

  http://hamelectronics.com/k1deu/pages/ham/transmitters/am/pages/universal_speech_amplifier.htm

Thanks John,

I am getting close to scrapping the Hi-Z on the 12AX7. I still have more hum than I should. I do not see how a 4.7 or a 10Meg resistor is ever going to work in the Apache the way the mic input connector is so far away from the grid of the tube.

The remoted MIKE GAIN pot wiring is not so sensitive but boy that input is.

If you want to run the D-104 with a High Z load of 10 Meg, then in my opinion, that has to be done outside. I am even entertaining a small board located right at the Mic connector with an MPF-102 source follower on it. This would allow me to drive cable at a lower Z and put a 270K resistor on the grid of the 12AX7.

Mike WU2D

Mike; Its nice to have a JFET impedance converter in the base of a D-104 to force the mike input to a low impedance for less hum pick-up. The worst schematic ever http://hamelectronics.com/k1deu/pages/ham/transmitters/am/pages/poor_mans_eq.html     But

    In all transmitters I use new braid RG-58A/U from mike in jack to first audio grid. Also make sure you do not have ripple on any PTT parallel line. Next, in All tube transmitters, I use balanced filament wiring to the first speech amp tube usually a AX7 or AU6.
 This makes a night and day difference in High impedance mike hum pickup. Do the standard stuff. Twist two equal small insulated conductors together to run from New floating (at tube) filament pins (neither grounded to chassis) to the filament transformer winding where one side is grounded to chassis (OK, fine, irwin !). No current is required to light one tube so use tiny wire. Now we cannot have a ground loop bringing 60 Cycle hum directly up our grid leak resistor ! This great technique came from Uncle, Peter W1VZR many years ago. 73  John   

2 years ago, I converted a DX100 from 1625s to using EL34s, with good success. I also tested 6146s and found that they  required a lower screen voltage, which made my job harder, and were also too fat for good cooling in that space. Not to mention the wide sockets. You can read my reports here - written over a period of weeks:
http://amfone.net/Amforum/index.php?topic=3789.0

IMHO, the EL34s were perfect for the Apache-sized transmitter with a pair of 6146, and DX100 of course.