Heathkit HW-8 Information
>From: Mike.Czuhajewski@hambbs.wb3ffv.ampr.org (Mike Czuhajewski)
>Subject: More core info; long one
>Date: Thu, 17 Nov 94 23:52:14 EST5EDT
17 Nov 1994--Here's some more info from the Idea Exchange in the QRP Quarterly dealing with bad HW-8 cores. This is all mine, except for one portion written by Dave Benson, NN1G.
--WA8MCQ
From the April 1993 Idea Exchange in the QRP Quarterly--
The October 1992 issue of the Quarterly contained my article on the output cores going bad and lowering power on 80 and/or 40 meters. I included the results of some tests of the FT37-63 coils, both good and bad, on a Q meter. Since that time I fixed the 8th HW-8 with this problem (see, it's NOT an isolated incident!), and did some tests on the FT50A-63 coil. (That's L27, the larger of the two 80 meter coils.)
Nominal value of L27, per HW-8 manual: 27.5 uH
Bad L27, measured at 2.5 MHz: 35.2 uH, Q 220
Good L27, wound on fresh FT50A-67 core: 27.0 uH, Q 360
The new coil was wound with the same number of turns as the original. The inductance came down to the correct value, and there was a dramatic increase in Q. (Measurements were done at 2.5 MHz, one of the frequencies at which inductance can be read directly off the dial on the Boonton 260A Q meter.)
The 40 meter cores from a previous HW-8 both measured 10.1 uH at 7.9 MHz (another "standard" frequency on the Boonton) with Q of 186 and 196. Coils wound on fresh FT37-67 cores were trimmed to the nominal 7.0 uH, and their Q values were 300 and 337. (Although the original core material for the 80 and 40 meter coils is type 63, type 67 is a replacement for it and may be used.)
>From the July 1993 Idea Exchange--
COMMENTS ON WA8MCQ'S "BAD HW-8 CORES"
From Dave Benson, NN1G, our technical editor--The changes in ferrite characteristics referred to in Mikes article, "Bad HW-8 Cores" stems from core overdrive. In the olden days, ferrite cores were deliberately driven to saturation to provide non-volatile storage (remember core memories?). Dad--tell us again what it was like in the Mesozoic era!
Figure 11 [not included here] shows the relationship between current through a core winding (H) and the resultant magnetic field density (B). In normal applications, the ferrite core starts its life at the origin "O". When operated out to the point P1, the core follows the path associated with P1 thereafter. The harder the core is driven, the closer to the corners (P3) the core operates, and the more the permeability is "permanently" shifted as the core adopts a new operating path. Permeability is the slope of this operating curve, so the ferrite has taken on a new effective value. The core isn't really damaged--it's just gone to live in a bad neighborhood! Unfortunately, the only way to get the core back to its birthplace is to bake it at high temperature. (Replacing the fool thing seems easier, somehow!)
I'd guesstimate that for a typical ferrite core a high current spike in the amperes range would be sufficient to cause the permeability to shift appreciably. Assuming your rig uses a high current power source like a storage battery, simply touching a grounded probe to the wrong point while the circuit is under power could cause this effect. (As a bonus, of course, you get smoke; this is known as the "Real Men Don't Use Fuses" school of design.) The lightning-induced surge damage that prompted Mikes investigation is also a very plausible cause for this phenomenon.
--de NN1G
DELIBERATELY ZAPPING SOME PERFECTLY GOOD CORES
From me, WA8MCQ--Naturally, I couldn't resist giving this one a try--deliberately hitting a core with a huge overload to see what would happen. I had a lot of fun and wrote a nice piece on my experiments, but then lost it. Through one of those freak accidents that happens sooner or later to everyone with a computer, about two weeks before my deadline for this issue I found out that the entire column had been obliterated from the main disk and the backup as well. Only two paragraphs remained, so here's a condensed version rewritten at the last minute, pieced together from memory and notes.
The basic idea was to wind several turns on various cores, charge a capacitor of tens of thousands of microfarads with a power supply, then short the coil across it. The capacitor would insure a healthy current spike. (This technique is from the "Hit 'Em with a Mack Truck Doin' 90" school of experimentation.) Inductances would be measured before and after. I tried a variety of voltages as high as 15 but eventually settled on 5V as my standard value, for no reason in particular; I found out it didn't make any difference if I used higher voltages. (It's like asking the death row inmate if he wants his electrocution to be done with 500 volts or 2000--the end result is the same either way.)
First, I took an already-bad FT50A-63 which came from an HW-8 output network, the eighth one I cured. I used it as-is, with the original wire still on it. Measurements were taken at 2.5 MHz on my Boonton 260A Q meter for this one; the results--
Bad core before zapping: 35.5 uH, Q 215
After zapping: 38.2 uH, Q 153
Fresh core, same # of turns: 27 uH, Q 360
Next, a good FT37-67 core, a type also used in the HW-8 output nets on both 80 and 40 meters. I put 19 turns of #24 wire on it, and measured 6.19 uH at 7.9 MHz, with a Q of 307. Zapping it with 5 volts from the cap made it jump to 10.97 uH, while the Q plummeted to 45! I cranked the voltage on the cap up to 15 but it didn't make the core any worse than it was. (This is the "Mack Truck vs. Freight Train Comparison.") I tried 15V again with reversed polarity, which made no change in the measurements, although the magnetic flux lines would be reversed. I tried zapping it several times with a lower voltage each time, with negative results.
How about an FT50-61? Interestingly, the inductance on this one went down for some reason, from 23.7 to 19.4 uH, instead of increasing like the others did. The Q dropped from 153 to 108. This was the only core tested which showed a decrease in inductance. An FT37-61 tested later showed the expected increase; I don't THINK I reversed the figures when I wrote them during the experiments, but a decrease sounds fishy.
I tried an FT50-43 with similar results. I couldn't measure the Q, since it wouldn't give a reading on the Boonton; I had to use a borrowed LCR meter instead, which measures inductance with a 1 KHz tone. Unfortunately I couldn't find my notes for this one, but it too showed a significant increase in inductance.
Is this phenomenon limited to ferrite cores with high permeability (ui)? Obviously not, since the type 63 (or 67) has ui of 40, type 61 is 125 and type 43 is 850. How about powdered irons? Their permeability is much lower, and the material is somewhat different. Conventional wisdom is that powdered irons return to their original value after overload is removed. (For example, see the W1FB toroid article in the June 1993 issue of CQ magazine, and my item in the Idea Exchange in July 1990, "Cooking With Toroids".)
Surprisingly, they exhibited the same permanent shift in inductance (and thus permeability) but to a much lesser extent. I checked type 2 (ui of 10) and type 6 (ui of 8) cores, and they had small but noticeable shifts in inductance and Q. These were done with various amounts of wire and at different frequencies--
T50-2: before, 1.74 uH; after, 1.77 uH
T68-2: before, 3.90 uH; after, 3.94 uH
T37-6: before, 0.725 uH, Q 207; after, 0.728 uH, Q 194
T50-6: before, 2.18 uH, Q 210; after, 2.21 uH, Q 208
My earlier experiments (July 1990 QRP Quarterly) were done with a good quality Hewlett Packard test unit, but I only had readout to tenths of a microhenry. The T37-2 cores which I had cooked read 1.8uH before and after brutalization. This time I used my Boonton 260AQ meter, which resonates inductance with a well calibrated variable capacitor, with direct readout to 0.1 pF, allowing small changes to be easily observed. The earlier cores had changed, but I was unable to observe it.
Finally, another experiment with the FT37-61 I zapped. It had started at 7.2 uH and Q 340, and was zapped to 12.7 uH and Q 45. Dave had suggested the possibility of restoring cores by cooking them at high temperatures. I desperately wanted to take it to work and run it through the furnace used for firing thick film hybrid substrates, but the clean room supervisor made it abundantly clear that I would NOT put ANY foreign materials in HIS oven. On to the low-tech approach...
While my wife was baking a casserole one night at 350 degrees F, I popped the core into the oven for about 20 minutes (with the wire removed, just in case the enamel insulation might start smoking and stink up supper!) The core then measured 11.27 uH and Q60; not a dramatic change, but measurable. Of course, I had removed the wire and then rewound it, so the turn to turn spacing was somewhat different but I kept it as close as possible. Next, I removed the wire again and slipped the core over the tip of a Weller W60 soldering iron with 700 degree tip. I let it cook for about a minute, then melted a bit of solder on the core to make sure it was good and warm. After it cooled and I rewound it, the Q had increased a hair or two and the inductance went up to 13.23 uH. At this point I cut it in half with wire cutters and tossed it in the trash; I'd never trust THAT core in any of my circuits!
I have type 0 cores (tan) in a few sizes, but didn't bother zapping them since type 0 material is physically incapable of changing permeability. Although it is usually lumped together with powdered iron, it is actually made of phenolic and contains no iron of any sort. Its permeability, 1, is the same as air. As a Micrometals engineer told me on the phone once, it has the same magnetic properties as a block of wood. The Micrometals catalog makes it clear that it is actually phenolic, and that is also mentioned, but well hidden, in the Amidon toroid book (not their "road map" brochure), on page 44 of the February 1992 edition.
The bottom line? Although I didn't do a great deal of experimentation and there wasn't a great deal of precision, a reasonable conclusion is that ferrites of type 63/67, 61 and 43 can be changed substantially by gross overloads, while type 6 and 2 powdered irons exhibit the same characteristic but to a much lesser extent. In fact, with the latter you probably couldn't tell whether a given core had been zapped unless you had previously measured it and had a basis for comparison. The variations I saw on those were well within the 5% inductance tolerance that Micrometals specifies for them. Although no other types of ferrites or powdered irons were tested (it was starting to get expensive), it is reasonable to assume that all would behave in the same way.