Heathkit HW-8 Information



Heathkit HW-8 Output Cores
Complements of the KB7RGG Nostalgic Kits Central Website
Converted to HTML 12-14-1998 by KB9JJA
Reprinted here with permission from the author

BAD HW-8 OUTPUT CORES

Michael A. Czuhajewski WA8MCQ
7945 Citadel Drive
Severn, MD  21144

Here's the "reprint" of my QRP Quarterly article on the subject. Unfortunately this isn't exactly as it appeared in the QRP Quarterly; the note at the head of the disk file says it isn't the final version sent in--that was done at work on a different computer and I no longer have the disk, but this has most of the info. (No graphics included here--sorry, I hate ASCII art!) I went back to the printed article and updated this, so it's reasonably complete. This originally appeared in the October 1992 QRP Q, and much-condensed versions later appears in QST Hints & Kinks and SPRAT (GQRP).

Note to WA7NTF, AB4EL and anyone else who is so inclined--you may put this article in your QRP files for FTPing or whatever.

--WA8MCQ, 17 Nov 1994


A couple years ago, W8KYD sent me his HW-8 for repair after a nearby lightning strike. Among other things, it had very low output on both 40 and 80 meters, a quarter and half watt respectively. I tried tuning everything up to no avail, followed by extensive troubleshooting over several weeks, and nothing made much sense. I even went so far as to put my own HW-8 beside it, and transmit with parts of both rigs--his VFO feeding my mixer, driver and amp, his mixer feeding me, etc. Everything kept pointing to the area of the final amplifier, and I finally gave up in desperation and replaced the toroid coils in the matching network, even though toroids never go bad. WRONG! The output shot up to normal, and I had to admit the unthinkable--the cores themselves were bad, something I would never have suspected. Toroid cores NEVER go bad! (Famous last words...)

The permeability of the cores had increased for some reason, perhaps from a surge due to the lightning strike, thus increasing the inductance. This in turn shifted the tuning ranges down below the bottom of the amateur bands; no problem if you have a license to transmit on, say, 3.1 or 6.7 MHz, but few of us have one of those! Permeability of both powdered iron and ferrite toroids will change if they are overloaded. However, when they cool, powdered irons will return to approximately the original value, while ferrites will not. (See the Idea Exchange in the July 1990 issue of the QRP Quarterly, "Cooking With Toroids", which reported some experiments I did along those lines.) The HW-8 happens to use ferrites on 80 and 40 meters, with powdered irons on 20 and 15, which may explain why the latter bands were still good.

When I put the 80 and 40 meter coils from my HW-8 into his rig, both bands came up to normal output. Next, I peeled off two turns each from his 40 meter coils to reduce the inductance and put them back in--the output went up a bit from the original quarter watt. I peeled off a few more and the output was even higher. However, as I peeled off turns, his cores started running warm to the touch.

THE CURE

I wound new coils on fresh cores of the proper type, with the original number of turns, and his rig then put out normal power on both 80 and 40. The output nets on both bands use type 63 ferrite, which seems to be relatively uncommon in ham use.; It has a permeability of 40, much lower than the commonly used 43 and 61 mixes (which are 850 and 125, respectively).

COILS IN HW-8 OUTPUT NETS:
 
Band  Coil  Nominal Inductance Core
80 L26 15.5 uh FT37-63
  L27 27.5 uh FT50A-63
40 L28 7 uh FT37-63
  L29 7 uh FT37-63

For those not familiar with the FT50A-63, it's the same as an FT50-63 except that it's taller. As with all "A" cores, the inner and outer diameters are the same as the "regular" cores of the same type, as is the permeability. However, the Al factor is greater due to the increased cross section of the "A" core.)

I later received an HW-8 from KM4ZH, another QRPer, which had low output on 80 meters only. Once again, it was a bad set of toroids in the output network, and some new ones fixed it right up. Interestingly, I did not get full power at first when I put in a fresh set. Wound with the same number of turns as the old one, the new L27 didn't allow much more than a watt output. I had to peel off several turns before it reached full power. Variations in permeability between nominally identical cores accounts for this, so be prepared to experiment a bit with the number of turns.

VERIFYING THE PROBLEM

Originally, I would have thought that bad toroid cores would be quite rare; however, as of June 1992, over two years into this, I have 7 confirmed cases (one DX and 6 domestic) of low power on 80 and/or 40 being cured by new toroids. It's almost a religious quest with me now; every time I hear of someone locally who has an HW-8, I offer to check it out for them and tune it up, just so I can get the chance to search for more bad cores to prove that it's not a fluke. (By the way, I've checked at least 15 HW-8s and none had low power on 20 and 15 attributable to bad cores.)

If you have low power on either of these bands, try all the usual things first, such as aligning the rig, checking for shorts and opens on the circuit board, cleaning the bandswitch contacts with spray cleaner, etc. If it doesn't respond to these, then suspect bad toroids in the amplifier output.

To confirm your suspicions, put an RF probe or scope on the collector of the 2N4427 final amp (Q9) and read the voltage while transmitting into a dummy load. If it is substantially lower on the band(s) with the low output, then the problem is somewhere before the final amp--there isn't enough drive. If the voltage is about the same as the other bands, yet the output is quite a bit lower, give the coils the evil eye--the power is being generated but isn't making it past the network.
 

Peak voltages seen in bad HW-8:

Band Collector of Q9 Output to 50 Ohm Load Output Power (Watts)
80 18 7.06 0.5
40 18 4.9 0.24
20 18 12.09 1.46
15 14 11.76 1.38

Putting the coils back on the board is a real pain, since the tuning capacitor gets in the way. You can take the front panel off to remove the cap, but here's a better method. Take two pieces of very small wire about a foot long. Stick the ends through the holes, from the foil side of the board, and push up past the components. Solder the coil leads onto them with a lap joint, and pull it down into place. (You may need to enlarge the holes slightly for the solder junctions to pass.)

BIG VOLTS IN THE LITTLE GREEN BOX

Unlike newer QRP transmitter designs, the HW-8 does not use a low pass filter after the transistor; it uses the output network shown in Figure 1. [not included in this electronic article] (The input capacitor to ground is only used on 80 and 40 meters.)

Surprisingly high voltages are present within the network, which has some high impedance points. For example, with a good HW-8 running well over 1.5 watts output to a dummy load I checked the voltages at various points, using a Tektronix 465B scope and X10 probe and saw the peak to peak voltages below. (It was necessary to retune for maximum output every time the probe was moved, due to its small but finite capacitance affecting the network.) These are similar to those seen in other HW-8s on these bands.

Yes, that really IS four hundred and ten volts peak to peak on 80 meters, or 205 volts peak and 145 volts RMS. Nonbelievers are invited to verify this for themselves with a high impedance probe. Lacking that, you can touch one lead of a neon bulb to the circuit board and watch it light up. The other voltages may not be enough to fire a neon, depending on the particular bulb used. During one test, a 22K resistor placed from Point D [rotor of front panel transmit tuning cap] to ground reduced the voltage at the antenna connector from 28 volts P-P to 22 volts.
 

PEAK TO PEAK VOLTAGES SEEN IN HW-8 OUTPUT NETWORK

Band A B C D E
80 37 144 87 410 28
40 34 116 58 160 26

[Point A is collector of final amp; B is at the left side of the left coil in the net; C is the junction of the two coils; D is the right side of the right coil, also the rotor of the front panel cap; E is the stator of the tuning cap, which also appears at the antenna connector]

Voltages were lower on the other bands and not as impressive, but still rather high. Note that the highest voltage is at Point D, which is the rotor of the loading capacitor. If you take its knob off, you'll see that the last half inch of the shaft is plastic, not metal, for safety. Be careful where you put your fingers while transmitting with the covers off. You may not get fried, at this power level, but I'm sure you'll feel it!

For comparison, I checked a kit version of the W7EL Optimized QRP Transceiver (from Small Parts Center, which is no longer in business). That rig uses a 5-pole low pass filter between the output transistor (2N3553) and antenna connector. Running close to 2 watts output on 40 meters, I saw about 28 volts peak to peak at the collector of the 2N3553, in the center of the filter, and at the output.

USE THE RIGHT TYPE CORE!

You may make one substitute for the type 63 material--type 67 has the same permeability and is an improved mix, and is replacing type 63. Otherwise, do not make substitutes. Since I was too lazy at first to order the type 63/67 cores, I tried an FT37-61 since I had some on hand. It has higher permeability, and I calculated that only 11 turns were needed to give the necessary 7 microhenries for L29 (40 meters). For L28, I used the good coil from my own HW-8. Although it had the proper inductance, the -61 core only gave about a quarter watt output, and after 10 seconds it was almost too hot to touch! Obviously it was not the proper mix for this application. Avoid the temptation to use what's already on hand, and take the trouble to get the right cores.

I compared some bad cores with fresh ones, using a Hewlett-Packard 4276A LCZ meter at a test frequency of 1 KHz. (The number of turns used here is different from those used in the HW-8.)

FT37-63 cores, all 20 turns #28 wire:
     Core 1, bad, 9.9 uh
     Core 2, bad, 9.0 uh
     Core 3, bad, 10.0 uh
     Core 4, fresh, 5.7 uh
FT37-63 cores, both 15 turns #28 wire:
     Core 5, bad, 5.5 uh
     Core 6, fresh, 3.2 uh
FT50A-63 cores, all 20 turns #26 wire
     Core 7, bad, 13.0 uh
     Core 8, bad, 12.5 uh
     Core 9, fresh, 10.5 uh
     Core 10, fresh, 10.3 uh
 

When I fixed HW-8 number 6, I put the old coils on one of my Boonton 260A Q meters to check their specs. The nominal value for the 40 meter coils is 7 uh each; both of these coils measured about 10.1 uh, with Q values of 186 and 196 at 7.9 MHz. I pulled a few turns off each one to reduce them to approximately 7 uh, and the Q values stayed about the same. By comparison, a pair of new coils wound on fresh FT37-67 cores and trimmed to 7.0 uh had Q values of 300 and 337. (I checked them at 7.9 MHz since that is one of the "standard" frequencies on the 260A at which inductance can be read directly from a scale on the variable capacitor dial.)

As I mentioned earlier, there are many things that can cause low output on 80 and/or 40, and bad toroids are only one of them. However, having had 7 confirmed cases so far I think it's safe to say that it's not all that rare. If you have the output problem and all else fails, don't put on a fake-nose-and-eyeglass disguise and sell your beloved HW-8 at the next hamfest--try replacing the cores first and you might get a pleasant surprise.
 
--QRP--

Michael A. Czuhajewski WA8MCQ
7945 Citadel Drive
Severn, MD 21144

Last update: 7 Sep. 2002