The AM Forum
March 28, 2024, 11:16:01 AM *
Welcome, Guest. Please login or register.

Login with username, password and session length
 
   Home   Help Calendar Links Staff List Gallery Login Register  
Pages: 1 ... 10 [11] 12   Go Down
  Print  
Author Topic: starting work on the amp  (Read 185926 times)
0 Members and 1 Guest are viewing this topic.
Opcom
Patrick J. / KD5OEI
Contributing
Member
*
Offline Offline

Posts: 8308



WWW
« Reply #250 on: April 05, 2017, 07:27:53 PM »

Never tried the fast diode. What diode could switch at 30MHz much less 50MHz and handle the current? The peak current would possible be a few amps.

I tried a 5Y3 in the cathode circuit of a 3-1000 a long time ago. It seemed to work fine but was driven by a 2x 6146 amp and its pi-network output, so I can't say how well the 5Y3 worked.

The current in the cathode circuit of a large GG tube amp looks like 1-2A if you take the drive wattage and the impedance and do the I2R math, but I'm sure that's not very accurate because peak values can be much more and it's a half wave.

A tube with a resistance of 50 Ohms is not so easily found, or is it?

Maybe something to throw on the table could be a high current low impedance triode like the 6AS7/6080, 6336, or even the monstrous 7242.
Plate resistance:
6AS7/6080 = 280 Ohms
6336 = 200 Ohms
7242 = 82 Ohms

I'd like to explore a tube or diode.

I'm leaning toward a low Q tuned circuit as is traditionally used in GG amps but have little experience in choosing values for that.

* 7242.pdf (363.9 KB - downloaded 420 times.)
Logged

Radio Candelstein - Flagship Station of the NRK Radio Network.
WBear2GCR
Member

Offline Offline

Posts: 4135


Brrrr- it's cold in the shack! Fire up the BIG RIG


WWW
« Reply #251 on: April 05, 2017, 09:11:30 PM »

Good advice on the ferrite beads! There are many strange things going on with that coil experiment and the picture may help explain, the lash up is very something. The coil and 4-400 chassis/power is not shown in the picture. A 2x 811 chassis in the first picture is being used as a test pig.


<snip>


I'd avoid ferrite beads in this situation.

Small resistors, about 100ohms as close as possible to the actual gate leads. Aka "gate stoppers".
Then small pf caps from the source to gate. Sometimes pf + resistor.

Should kill the unwanted HF or VHF parasitics.

Just reading backwards now...
Logged

_-_- bear WB2GCR                   http://www.bearlabs.com
WD5JKO
Member

Offline Offline

Posts: 1996


WD5JKO


« Reply #252 on: April 05, 2017, 11:15:29 PM »

I tried a 5Y3 in the cathode circuit of a 3-1000 a long time ago. It seemed to work fine but was driven by a 2x 6146 amp and its pi-network output, so I can't say how well the 5Y3 worked.

   Would sure be interesting to do a 2 tone test and try something like that. A lower impedance diode would be a damper diode like a 6AX4, or maybe 2 in parallel. Looking at the diode curve, at 70v it will pull around 800ma, so R=E/I = 88 ohms. The impedance will be dynamic with drive level, just like it is with the G-G RF tube.

Jim
Wd5JKO

* 6AX4GTB.pdf (168.16 KB - downloaded 405 times.)
Logged
Opcom
Patrick J. / KD5OEI
Contributing
Member
*
Offline Offline

Posts: 8308



WWW
« Reply #253 on: April 06, 2017, 11:26:00 PM »

Good advice on the ferrite beads! There are many strange things going on with that coil experiment and the picture may help explain, the lash up is very something. The coil and 4-400 chassis/power is not shown in the picture. A 2x 811 chassis in the first picture is being used as a test pig.


<snip>


I'd avoid ferrite beads in this situation.

Small resistors, about 100ohms as close as possible to the actual gate leads. Aka "gate stoppers".
Then small pf caps from the source to gate. Sometimes pf + resistor.

Should kill the unwanted HF or VHF parasitics.

Just reading backwards now...


Perfect! as my friend Henry SWL also needs to know this stuff since several FETs were sacrificed!
Logged

Radio Candelstein - Flagship Station of the NRK Radio Network.
Opcom
Patrick J. / KD5OEI
Contributing
Member
*
Offline Offline

Posts: 8308



WWW
« Reply #254 on: April 06, 2017, 11:44:30 PM »

I tried a 5Y3 in the cathode circuit of a 3-1000 a long time ago. It seemed to work fine but was driven by a 2x 6146 amp and its pi-network output, so I can't say how well the 5Y3 worked.

   Would sure be interesting to do a 2 tone test and try something like that. A lower impedance diode would be a damper diode like a 6AX4, or maybe 2 in parallel. Looking at the diode curve, at 70v it will pull around 800ma, so R=E/I = 88 ohms. The impedance will be dynamic with drive level, just like it is with the G-G RF tube.

Jim
Wd5JKO

That's a great idea! It would seem to fit the bill. What about the plate dissipation? I don't recall the 5Y3 ever getting red in the face but it's a higher impedance job.
Logged

Radio Candelstein - Flagship Station of the NRK Radio Network.
Opcom
Patrick J. / KD5OEI
Contributing
Member
*
Offline Offline

Posts: 8308



WWW
« Reply #255 on: April 09, 2017, 01:52:53 PM »

Two things need to be done.

1.) come up with a nice flywheel for the cathode circuit.

2.) look much more closely at the GPT-10K tank coil - because it's from something else. I harbor no fantasies that it will be plug and play. This seems to be the harder part so I have been working on that.

The next post shows what was found out about using the GPT-10K coil with the 3CX3000 and offers some options to consider.
Logged

Radio Candelstein - Flagship Station of the NRK Radio Network.
Opcom
Patrick J. / KD5OEI
Contributing
Member
*
Offline Offline

Posts: 8308



WWW
« Reply #256 on: April 09, 2017, 02:06:21 PM »

To re-introduce it, the big beautiful general coverage GPT-10 tank coil is made to last!

The coil has an integral bandswitch and is designed for 2-28MHz and its inductance in uH for each of its 9 bands in are:

13.750 - 2-3 MHz
9.750 - 3-4 MHz
5.850 - 4-6 MHz
3.750 - 6-8 MHz
2.250 - 8-11 MHz
1.500 - 11-15 MHz
0.950 - 15-19 MHz
0.600 - 19-24 MHz
0.225 - 24-28 MHz

Now I have to point out that this may contain all kinds of errors, so don't take my word for it. The big point discussed in here is about the Q of the plate circuit.

It was designed for the 4CX5000 operated in GG mode but with normal electrode operating conditions G2=1200V, G1=negative 300V, running 7500V @ 1.5A.

This alone, just looking at an E/I result gives 5000 Ohms. Not the plate load but a comparison figure.

(The Eimac data calls for AB1 and 1.9A plate current @ 10KW output, but the GPT-10K uses cathode drive and 1.5A (1.75A max) plate current. The Eimac data calls for 1.9A but it makes no sense to quibble when the equipment manual gives application data.)

The amplifier application here is the 3CX3000 in GG mode running 4800V @ 1.7A.
The dividend of these values gives a figure of about 2800 Ohms.

From this, it is easy to see that for an equivalent L/C ratio or "Q", the plate coil from the GPT-10K has inductance values that are quite higher than those desirable for the 3CX3000 application.

I believe that tediously calculated values for actual plate load impedance in both cases would further bear this out so I skipped it for now.

For this thought-experiment, which should become a real experiment, the lowest band of interest, 2-3MHz, is considered. It also needs to be expanded down to 1.7MHz to cover 160M.

The Q will vary a lot over this range and that is why this band is a good choice to use in order to consider how to use the coil.

Using the PI-EL program, which seems as accurate as any other, some "Q" numbers on the GPT-10K bandswitched plate coil, 3CX3000, and 50 Ohm output are:

1.7MHz: 15
3MHz: 7.8

If a Q  of 7.8 is not acceptable, then the inductance of 13.75uH is too large. If the bandswitch tap were moved down one position to 9.75uH, these "Q" numbers are obtained:

1.7Mhz: 22.2
3MHz: 12.1

It looks like a high Q is probably acceptable for this coil because it is made 1/2" silver plated copper tubing and it will not need to handle 10KW, but by the limits of the tube, only about 6KW. The equivalent coil current decrease should be:

1-(SQRT 0.6) or 22.5%.

Going back to the complete set of PI-EL calculations done with this coil's series of inductances (post later), the Q seemingly improves as the frequency goes up. I didn't try to investigate this but it is interesting.

Thoughts on ways to use the coil including re-tapping the inductance as above:

1.) re-tap the inductance to values from calculations so that on each band Q=>10 is obtained

2.) Use a higher plate voltage on the 3CX3000 but the tube is rated 5KV absolute maximum and it would violate design rules.

3.) Run a lower current at "full output", meaning higher load resistance and much lower output power target. It could more closely match the original application of the coil if the maximum plate current was limited to 1A.  Output might be limited to 3500W. (the horror!) and presumably only 170W of drive to make the 3500W. This is not a bad thing but not good either as a 3500W rating is extreme stupid underkill for what is being built considering the equipment size and power supply capacity.

4.) Design pi-network for output impedance of 200 Ohms and use the intended balanced tuner to handle the ladder line (or even a correctly sized 2:1 transformer to get to 50 Ohms if that's wanted) Here is the data calculated for 50 ohms vs 200 Ohms:

50 Ohm output
1.7MHz: Q=15, Cin=732pF, L=13.75uH, Cout=3781pF
3MHz: Q=7.8, Cin=221pF, L=13.75uH, Cout=767pF

200 Ohm output
1.7MHz: Q=21, Cin=845pF, L=13.75uH, Cout=2440pF
3MHz: Q=11.5, Cin=267pF, L=13.75uH, Cout=743pF

This would be quite a transformer but there is precedent in the V.R. Stokes book and if I am correct, a transformer just like a balun works well if designed for the frequency range and impedances. Hmmm..
Logged

Radio Candelstein - Flagship Station of the NRK Radio Network.
KD6VXI
Contributing
Member
*
Offline Offline

Posts: 2648


Making AM GREAT Again!


« Reply #257 on: April 09, 2017, 04:31:41 PM »

Pat,

I can tell you from recent experience, the 3cx3000 will take 6kv plus on it.  The xformer here was a 4kv AC output into a fwb, cap in.  That's 6400 DC.

15 reverse connected 10A10 diodes brought the ZSAC close to what a 3x3 on 4.8 kv was.  I then added a variable bias adjustment, for good measure.

If you want to bump your plate voltage up you'll be OK up to the mid 6kv range.  Some of the CB mobiles had 10 kv plus.  A lot of bias, but the tubes didn't arc.

As to Q, I like low Q.  7 to 8 on the output, 2 on the input.

--Shane
KD6VXI
Logged
Opcom
Patrick J. / KD5OEI
Contributing
Member
*
Offline Offline

Posts: 8308



WWW
« Reply #258 on: April 14, 2017, 05:35:32 PM »

Shane, those are very good points.

The HV is possibly able to go to 6KV by using the right taps.  One thing that caused my last post/question (a complicated question indeed!) was that I found the Q of the tuned circuit, with 1720 Ohms load to the anode and 50 Ohms on the output, would look good at one end of the band and rotten at the other. I guess I knew that was coming but did not realize how large a difference there could be.

On the other hand, Q of 7 or 8 might be OK, but am I mistaken to worry about how clean or efficient the AB2 amp will be with a low Q? I also worry about excessive Q causing a high frequency rol=off of the audio. I would not want the amplifier to make me down 6dB@4KHz due to wasting power by part of the signal being off resonance.. Audio bandwidth shaping, if desired, should be done in the audio control amplifier and EQ. I do not know the math on calculating that but will have to look into it.

Back to the PI network investigation, I created a spreadsheet, taking knowledge from the typical formulas in the ARRL Hand book and Radio Handbook, which agreed.
It turns out that the various online calculators and downloadable programs are not all created equal, do not all agree, and they demand that the user 'do it their way'. This is fine for straightforward building, but I wanted to learn more by plugging in a Q value, input Z and output Z, then having a spreadsheet calculate the required input capacitance, inductance, and output capacitance. This took me a couple three hours to get it right and double check it.

So, I am halfway there, with a spreadsheet that takes in plate load, output load, and Q value; and gives C/L/C for 100KHz increments from 1.5 to 60MHz. There are limits to it - very low Q values mixed with high plate loads give errors. It's probably the square root stuff and limits of the excel format. Maybe someone can figure out why.

What's wanted is to add a sheet to take inputs of L value, input Z and output Z, and from that provide the Cin, Cout, and Q. This is for those who are determined to make an existing coil work.

This file has been saved as .xls and .ods and here the xls version is, so others can participate in this numerical experiment. No macros, nothing hidden. Those things make a mess in the computer and in the mind.

Others can try it. I hope it is fun and I accept responsibility for any errors, just please mention them to me so I can fix them.


* calculator for tube pi network_v2017-04-14-1619.xls (416.5 KB - downloaded 448 times.)
Logged

Radio Candelstein - Flagship Station of the NRK Radio Network.
KD6VXI
Contributing
Member
*
Offline Offline

Posts: 2648


Making AM GREAT Again!


« Reply #259 on: April 14, 2017, 07:54:53 PM »

I feel you on the different calculators giving different outputs.

I've found that Tonne's and Ian G3SEK spreadsheet give the best.  With Tonne's software, you have to take stray C, Lsupp, etc into account after the numbers are spit out.  Ian's spreadsheet, on the other hand, will darn near give you exact values as long as you give it all the data it asks for.  Ie, on the 3cx3000 just built, Tonne's program gave  about 28 pf for Ctune.  After running through Ian's spreadsheet I found I needed about 8 pf Cmin to get to 29.7 .  That necessitated getting a new cap for Ctune.  Prior I couldn't tune about the tech and of ten.

You give up some harmonic cleanliness with a low q.  You gain bandwidth excursions without returning (your pass and is wider) and you also usually have a lot less peaky tuning.  A Q of 2-3 is what's regarded as OK on inputs, and that lets us hopefully tune the entire band under 1.5 or so.  But, there is very little harmonic suppression compared to a Q of 8.  And when you go up to 12, it's even better.

What I'm not real sure about is IMD in relation to tank Q.  Maybe Tom, K1JJ could chime in on that.  I'm sure he's mentioned it before, but I plead CRS now :-)

--Shane
KD6VXI
Logged
Opcom
Patrick J. / KD5OEI
Contributing
Member
*
Offline Offline

Posts: 8308



WWW
« Reply #260 on: April 15, 2017, 11:58:14 AM »

It's said the pi network is made of two L networks, as in the figure. My main difficulty is the messy equation indicated.


* L networks.png (42.17 KB, 1123x508 - viewed 848 times.)
Logged

Radio Candelstein - Flagship Station of the NRK Radio Network.
K1JJ
Contributing
Member
*
Offline Offline

Posts: 8893


"Let's go kayaking, Tommy!" - Yaz


« Reply #261 on: April 15, 2017, 03:07:17 PM »


What I'm not real sure about is IMD in relation to tank Q.  Maybe Tom, K1JJ could chime in on that.  I'm sure he's mentioned it before, but I plead CRS now :-)

--Shane
KD6VXI

Hi Shane,

Generally, the higher the Q, the better the IMD figures, but the lower the efficiency.  To take an extreme case, just imagine a plate tank Q of 2 - hardly any flywheel effect to integrate the class B pulses.  A  Q of 12 is a decent compromise since efficiency is important to most hams. But if pure cleanliness and great IMD figures are most important, then:

I proved this out to myself a few years ago when building and testing my super linear amplifier chain.  I started with -75 dB 3rd order IMD from the 100mW output of the FT-1000D > 1 watt precision lab amp > 3CX-350J > 8877  > pair of 8877s.

The class A 3CX350J ran a plate tank Q of about 50.  The  8877 class A  driver ran a Q of about 30 IIRC.  I was able to get an IMD of about  -55 dB 3rd from the chain's output after tweaking everything and running it at ~ 1/2 power from max. The high Q and class A elements were most key to this excellent IMD performance.  Super clean drivers are so important. They can make or break a system.

Of course, these are old school techniques. An SDR using pre-distortion ("Pure Signal", etc)  would make all of this futile. A big, "CB-quality" solid state driver into the pair of 8877s would make more power out and still be -55 dB third order achievable....sigh.

T
Logged

Use an "AM Courtesy Filter" to limit transmit audio bandwidth  +-4.5 KHz, +-6.0 KHz or +-8.0 KHz when needed.  Easily done in DSP.

Wise Words : "I'm as old as I've ever been... and I'm as young as I'll ever be."

There's nothing like an old dog.
KD6VXI
Contributing
Member
*
Offline Offline

Posts: 2648


Making AM GREAT Again!


« Reply #262 on: April 15, 2017, 10:31:00 PM »

As always, no free lunch.

How about adding the L, as in Pi-L...  I'd imagine same / same, since you're adding more filtering.

Predistortion....  Yeah.  I'm currently contemplating a motorcycle or ANAN... Lol.  That's ham dedication there.

--Shane
KD6VXI
Logged
DMOD
AC0OB - A Place where Thermionic Emitters Rule!
Contributing
Member
*
Offline Offline

Posts: 1768


« Reply #263 on: April 16, 2017, 11:00:25 PM »

Pat,

If you have a calculator or MatLab you can use these equations:

Calcs. for Tube Circuits
clc
F = 3.7e6;
twopi = 2*(355/113);
twopiF = twopi*F;
% Tube Calcs
Ip = (Value of Plate Current here);
Vp = (Value of Plate Voltage here);
Pin = Ip*Vp
Vsat = 0.075*Vp;
Pout = (expected Power Output here);
Rout = 50;
Rplate = (Vp - Vsat)/(2*Ip)
Eff = Pout/Pin
TubeC = 15e-12;
%
% Wingfield Pi-Net Equation
Qz = 12;
Qzsq = Qz^2;
Q1 = ((Rplate*Qz) - sqrt(Rplate*Rout*Qzsq - (Rplate - Rout)^2))/(Rplate - Rout)
Q2 = Qz - Q1;
Xc1 = Rplate/Q1;
Xc2 = Rout/Q2;
XL = Rplate*Qz/(Q1^2 + 1);
CPT = 1/((twopi*F*Xc1)- TubeC)
L1 = XL/(twopi*F)
CANT = 1/(twopi*F*Xc2)
Logged

Charlie Eppes: Dad would be so happy if we married a doctor.
Don Eppes: Yeah, well, Dad would be happy if I married someone with a pulse.NUMB3RS   Smiley
Opcom
Patrick J. / KD5OEI
Contributing
Member
*
Offline Offline

Posts: 8308



WWW
« Reply #264 on: April 18, 2017, 11:35:06 PM »

some improvements.
I'll have to do some math later. Looks promising.

* _new-3x3K_plus6_v025.pdf (1095.61 KB - downloaded 508 times.)
Logged

Radio Candelstein - Flagship Station of the NRK Radio Network.
Opcom
Patrick J. / KD5OEI
Contributing
Member
*
Offline Offline

Posts: 8308



WWW
« Reply #265 on: April 19, 2017, 12:04:23 AM »

This is something I did Sunday, tedious, but after putting it into a table according to the coil and switch for the bands, it shows Q at the 'band edges' of the coil as-is.
The band edges that the coil was assigned can probably be changed.
Anyway the Q looks decent in most places so I think the coil will be OK.
Experiments should show how things need to be changed.
Pi-L is possible but it'll change everything.





* table of Q.png (65.29 KB, 1396x478 - viewed 783 times.)
Logged

Radio Candelstein - Flagship Station of the NRK Radio Network.
Opcom
Patrick J. / KD5OEI
Contributing
Member
*
Offline Offline

Posts: 8308



WWW
« Reply #266 on: April 22, 2017, 08:55:39 PM »

polished some copper today. For the tuning unit.

also worked on the steel panel that will mount the filament regulator


* 100_2401s.jpg (343.68 KB, 1000x767 - viewed 811 times.)
Logged

Radio Candelstein - Flagship Station of the NRK Radio Network.
Opcom
Patrick J. / KD5OEI
Contributing
Member
*
Offline Offline

Posts: 8308



WWW
« Reply #267 on: April 30, 2017, 08:18:53 PM »

The filament regulator is mounted. This is a 60 Lb 1KVA Sola unit. It has excess capacity and so the rectifier filaments will also be regulated. It is needed here because in the summer the mains in the shack with everything running HVAC etc. can vary from 210-240V

After the transformer A 3 Ohm 300W pot connected as a rheostat should let me get the voltage right on and allow +/- 0.15V fine adjustment and an expanded scale voltmeter setup will be used if possible. If the voltage can't be hit by choosing a tap on the transformer and adjusting the pot, then a few more Ohms may have to be added.
3CX3000 soft start: 700W
3cx3000: 390W
4) 673: 200W
-----------------
590W - an adequate load. 900W on soft start - the reson for the 1KVA rating.
The 673s don't need a soft start but the extra load on the transformer will help it run more efficiently.
This will have to be tweeked later.

So the panel is an old fashioned 5/32" steel panel for strength and is held in place with six grade 5, 1/4-28 bolts and nylock nuts. The transformer mounted with 3/8-24 bolts and grade 5 was also used there with nylocks and split washers.

The last three relays were also put in sockets. One is the filament inrush timer, the next is  for something, forgot right now, and the third is for keying and has a 24VDC coil. After having a mess of relays behind the rack due to the Tucker's non-isolated 120VAC keying circuit, going with a nice low voltage isolated circuit on this project looks attractive.

The plate supply soft start resistor was changed to 10 Ohms by putting the three resistors in parallel. It was 90 Ohms for experiments before.

So much still to do. the RF chassis will go back on the workbench but first a metal plate or set of rails has to be installed in the rack for it to sit upon and some sort of flexible hose has to be gotten and installed on the blower. Maybe dryer duct parts will work OK for that but I would prefer a smooth duct.


* 100_2439.JPG (395.11 KB, 991x1311 - viewed 873 times.)

* 100_2439as.jpg (346.33 KB, 1100x942 - viewed 773 times.)

* 100_2439bs.jpg (202.11 KB, 1000x484 - viewed 851 times.)
Logged

Radio Candelstein - Flagship Station of the NRK Radio Network.
KD6VXI
Contributing
Member
*
Offline Offline

Posts: 2648


Making AM GREAT Again!


« Reply #268 on: May 01, 2017, 01:04:56 AM »

If you worry about hitting the limit of the SOLA while on soft start, use the unregulated 240 through the resistors, and have the contactor switch to the regaled voltage.

The soft start doesn't need to be regulated.

--Shane
KD6VXI
Logged
Opcom
Patrick J. / KD5OEI
Contributing
Member
*
Offline Offline

Posts: 8308



WWW
« Reply #269 on: May 02, 2017, 01:48:17 AM »

I considered that, but there would be a short power cut at change over because the phase difference between the input and output sides of the regulator transformer would not allow a simple relay closure across a resistor, but a break then make of a few msec via heavy duty DPDT relay would be needed. The shock to the filament might be harsher with the interruption but still way better than just turning the filament on. The nice 10 ohm resistor in series with the filament transformer is also helpful for those with the Henry parts.  This might be nit picking but every little bit helps. I was reading that the structure of the tungsten changes as it passes through a certain temperature on its way to operating temp and that is is hard on the wire. Just want to be as easy and soft as possible. The article says that tungsten is subject to recrystallization as it reaches 1200 deg C so whether this only happens when the filament is first made or whether it happens 'slightly' each time the filament is heated is not explained there, but an easy does it approach should be best. anyway the attachments seem close to a gospel on it.

* tubes with thoriated tungsten filaments thoria05.pdf (2215.25 KB - downloaded 562 times.)
* tube-topics.pdf (1048.78 KB - downloaded 658 times.)
Logged

Radio Candelstein - Flagship Station of the NRK Radio Network.
Opcom
Patrick J. / KD5OEI
Contributing
Member
*
Offline Offline

Posts: 8308



WWW
« Reply #270 on: May 05, 2017, 02:01:20 AM »

http://amfone.net/Amforum/index.php?topic=42633.msg310468#msg310468 references the W4ZT bias board whic is popular.

I like that circuit but I have a complaint that it can only control current in one direction. I know I have seen it before and was puzzling over how well itworks yet is so simple. I do not have a TL431 device in LTspice, so could not examine this for myself.

When I have been doing bias supplies, I always wanted them to sink or source current.
A giant bleeder rated for 3-4x the 'bias current' in this case of the 3CX3000 amp is not practical.
This reminded me of an IC I used to mis-use.

The audio IC STK0050, a 50W 4 Ohm amplifier module from the 1980s that is very basic. It is really a few dice and a thin film 'ic' all assembled on an aluminum plate with a plastic cover. Anyway it is no longer available but the audiophiles of old Pioneer receivers have reverse engineered it and the schematic is published in a souped up version, so they make their own. Anyway I had made a few DC supplies, yoke drivers, and other things with them in the old days when they could be had for ten bucks.

The circuit can maintain a steady voltage with current flowing either way. OK well the cathode of the tube only lets current flow one way but the extra control is worth something to me. And like the W4ZT board or a string of diodes there are no giant bias bleeders. It may have other uses, such as when the screen current goes negative in some tubes, etc.

This is maybe more of just a thought experiment related to the project but the LTspice simulation looks very good so it seems worth sharing.

The setup is like for a 3CX3000. 4KV on the anode, 0-2A variation in cathode current at an audio rate 100Hz to make it easy to see.

Pictures:
1. the original STK0050 circuit (no values given) and the circuit from the audio message board, which someone made. It's a souped up version.

2. the circuit I was simulating.

3. the LTspice waveforms from that circuit.


schematic voltages:

V1 and V2 are just two 30V DC supplies in series. It's the STK0050 notmal kind of operationg voltage, except in this case the negative rail is grounded rather than the common tap.

V3 is a voltage that could come from a pot, with a range of 0 to 40V.

V4 simulates the 0-2A cathode current audio rate current by using a 4000V peak to peak signal biased 100V positive, through a 2000 Ohm resistor. This gives a brute force current waveform to be forced upon the bias supply just like wta comes off the cathode of the tube.

Waveforms:

V(output) is the bias voltage to the center tap of the filament transformer. In this simulation it is raised over time from about 1.5 to 38V then lowered back to near zero bias.

I(R1) is the current supplied by the upper amplifier the NPN stage. It varies from 0 to about 1A.

I(R12) is the cathode current through the filament transformer CT. -2A is the peak audio-rate signal, 0 is the trough of modulation, the negative peak.

I(R2) is the current sinked through the lower amplifier stage and varies from about 0 to -2A.

There are issues with this, for one thing the simulation does not take into account the reduction in cathode current signal amplitude as bias is raised or lowered. I don't see this as an invalidation but it is more like the drive was adjusted to make the same 2A peaks as the cathode bias was changed from zero bias to 38V.

OK so this may get off track but it seemed interesting enough to post.


* STK-0050_reverse eng.png (127.13 KB, 1261x533 - viewed 843 times.)

* circuit - Copy.png (35.39 KB, 1248x794 - viewed 778 times.)

* circuit waveforms - Copy.png (35.27 KB, 1269x890 - viewed 790 times.)
Logged

Radio Candelstein - Flagship Station of the NRK Radio Network.
Opcom
Patrick J. / KD5OEI
Contributing
Member
*
Offline Offline

Posts: 8308



WWW
« Reply #271 on: May 21, 2017, 08:49:27 PM »

Time to start fitting things together.


* IMG_20170521_144959a.jpg (146.36 KB, 941x1717 - viewed 910 times.)

* IMG_20170521_145027a.jpg (115.09 KB, 1256x860 - viewed 768 times.)

* IMG_20170521_145101a.jpg (61.02 KB, 864x825 - viewed 739 times.)
Logged

Radio Candelstein - Flagship Station of the NRK Radio Network.
Opcom
Patrick J. / KD5OEI
Contributing
Member
*
Offline Offline

Posts: 8308



WWW
« Reply #272 on: July 03, 2017, 09:44:48 PM »

Sorry for the delays. Parents and house maintenance issues been eating weekends constantly. Hah sometimes as slow as projects go here I think If I grow old and pass on before finishing this then someone will have to take up the job!

So the tube cooling has to be done. It takes up space and so will determine what else fits where. A few 'dryer vent' items were had from the hardware store. There's a round 6" to 4" reducer that can be cut and formed to fit over the square blower outlet and transition to 4" duct size. Screws will hold it to the blower and aluminum tape should cover any gaps. A flexible but somewhat rigid 4" duct (not the plastic with coil spring type or the thin foil-sided type, but the compressed aluminum tubing that holds its shape well) should get the air to the underside or back of the RF chassis.

The lower chassis is 4" tall and I hope I can feed the air in at the back rather than from underneath. If it has to come from underneath it will be a PITA because the HV rectifier stacks are under there and I don't want the aluminum duct in the same space.

A piano hinge on-hand will probably be used to join the front of the RF chassis together and two brackets with screws should take care of the rear. There should still be room for a tuned circuit inside. Once it is together the aluminum tape can seal around the crack between the two chassis as well as any holes. The filament wires from the transformer are too short, so some more #6 will have to be added.

We also want to look at the scopes and stuff gotten recently. (http://amfone.net/Amforum/index.php?topic=42929.msg311240#msg311240) At least the HP 122AR dual channel unit should replace the single channel 120AR on the bench meter stack for audio range and general power supply stuff. Henry SWL also wants a portable tube low frequency tube scope for Tesla coil tuning. Probably a DuMont 304 may do or HP 120. There may or may not be time for scope stuff. The amp will have to come first. The sticking point is the cooling arrangement, can't do more til that is figured out.

Whatever is done there will be a few more images posted.
Logged

Radio Candelstein - Flagship Station of the NRK Radio Network.
KD6VXI
Contributing
Member
*
Offline Offline

Posts: 2648


Making AM GREAT Again!


« Reply #273 on: July 03, 2017, 10:11:09 PM »

Your going to want to WAY oversize your blower Pat.

The corrugations in the dryer vent will kill airflow.

Much better to use pvc pipe as much as possible.  Smooth wall.  Will cut turbulence down a LOT.

Turbulence = slows airflow, causes blower to work harder and doesn't deliver as much pressure.  Also, will make a lot more noise.

FYI


--Shane
KD6VXI
Logged
Opcom
Patrick J. / KD5OEI
Contributing
Member
*
Offline Offline

Posts: 8308



WWW
« Reply #274 on: July 04, 2017, 10:12:55 PM »

this blower's last job was a 4CX5000 and a 4-1000 in a commercial transmitter. The round part of the outlet there is 4".  It's a 3450RPM capacitor run motor. I measured 1.75in s.p. H2O from it. Hope it is enough.

It's not the little blower that comes in those Henry amps, I think it should be enough. The two I have from Henry 2000W models both tested out at only 0.65" S.P. H2O. Enough to blast directly through the socket and remove 2KW heat but not much else.  IMHO those are really only good for a 4-1000 etc.

Anyway I will be putting a Dwyer Magnehelic gauge there sooner or later. Once it's plumbed up I can check the pressure drop in the air duct.

PVC is great but it's not at all a straight run, kinda horrible really and I am trying to figure that out.

choice 1. run the air in from the bottom, making the RF chassis a pain to ever remove.

choice 2. suffer a couple elbows and restrictions and plumb in the air into the cathode compartment from the back of the chassis.


* Blower_de_4XC5000_y_4-1000_TXr.PNG (2144.69 KB, 1852x1518 - viewed 820 times.)
Logged

Radio Candelstein - Flagship Station of the NRK Radio Network.
Pages: 1 ... 10 [11] 12   Go Up
  Print  
 
Jump to:  

AMfone - Dedicated to Amplitude Modulation on the Amateur Radio Bands
 AMfone © 2001-2015
Powered by SMF 1.1.21 | SMF © 2015, Simple Machines
Page created in 0.076 seconds with 19 queries.