From owner-qrp-l@Lehigh.EDU Mon Oct 6 20:49:47 1997 Received: from fidoii.CC.lehigh.EDU (fidoii.CC.lehigh.EDU [128.180.1.4]) by oucsace.cs.ohiou.edu (8.8.5/8.8.5) with ESMTP id UAA00517 for ; Mon, 6 Oct 1997 20:49:45 -0400 (EDT) Received: from Lehigh.EDU ([127.0.0.1]) by fidoii.cc.Lehigh.EDU with SMTP id <34937-35114>; Mon, 6 Oct 1997 20:47:46 -0400 Received: from nss2.CC.Lehigh.EDU ([128.180.1.26]) by fidoii.cc.Lehigh.EDU with ESMTP id <35056-35114>; Mon, 6 Oct 1997 20:24:07 -0400 Received: from m7.boston.juno.com (m7.boston.juno.com [205.231.100.196]) by nss2.CC.Lehigh.EDU (8.8.5/8.8.5) with ESMTP id UAA259755 for ; Mon, 6 Oct 1997 20:22:04 -0400 Received: (from gsurrency@juno.com) by m7.boston.juno.com (queuemail) id U]M01180; Mon, 06 Oct 1997 20:20:06 EDT Message-Id: <19971006.171710.6566.5.gsurrency@juno.com> Date: Mon, 6 Oct 1997 17:16:32 -0700 Reply-To: gsurrency@juno.com Sender: owner-qrp-l@Lehigh.EDU Precedence: bulk From: gsurrency@juno.com (Gary L L Surrency) To: "Low Power Amateur Radio Discussion" Subject: HW-9: The (nearly) complete guide to the mods. Part 1 of 3 X-Mailer: Juno 1.38 X-Juno-Line-Breaks: 0-3,8-9,12-13,15-16,21-22,27-30,40-41,51-54,60-61, 70-71,77-78,84-87,93-94,101-102,109-112,118-119,125-126,138-139, 145-146,153-156,162-163,169-170,174-175,184-185,192-193,203-204, 211-212,216-217,228-229,238-246 X-Listprocessor-Version: 8.1 beta -- ListProcessor(tm) by CREN Status: RO ******************************************************************* HIT DELETE NOW IF NOT INTERESTED IN HW-9 MODS ******************************************************************** The HW-9 is a fine rig. However, it *does* have some problems as delivered by Heath. In this article, I hope to show how to improve the performance of the HW-9 and encourage others to try their own hand at problems areas that their rig may exhibit. Feel free to contact me with your own mods, or any questions that this material may create. Since this mod list is quite long, I have broken it up into three sections so as to not cause some mailers to choke. If anyone wants it all in one piece, I can email it to you complete. As mentioned in a previous posting, I will address a number of problems that my HW-9 had. Please note that I find Molex IC pins to be invaluable when substituting components, as they can be individually installed for transistor, diode, capacitor, or resistor sockets. This saves damage to the PCB, and allows me to try different devices without reaching for the soldering iron each time, or removing or replacing the PCB between tests. These socket pins come on a "tree" carrier that you can use for various IC pin counts, and are separated normally by snapping off the "tree" after installation. When used individually, they are a little more difficult to get installed straight without burning your fingers. I use a 1/4 watt resistor as a holding tool as I solder the individual pins. VOLTAGE REGULATOR Change U402 from a 78L08 to a 78L09 regulator, and replace D409 with a jumper wire. This will give you a nice solid 9 volt regulated supply with better regulation and slightly cooler operation of U402, as it isn't dropping as much voltage. I thought I could hear a very slight "chirp" on the transmitted signal before I made this change, and the original setup only provided about 8.6 volt with the diode in series with the regulator's ground pin. The diode is *not* in the error feedback loop of U403, so the regulation is not what it should be. The additional few tenths of a volt and better regulation are welcome in the VFO, second IF filter / amplifier, and AGC circuits. Notice that as Heath designed the radio, the BFO isn't operating from a steady regulated source. Many critical circuits in the rig are powered direct from the poorly regulated 12 volt main supply, rather than an accurately regulated source. I will later describe mods to the PSA-9 power supply to correct poor regulation that can also contribute to frequency shifts during transmit. If you use a different power supply to run the radio, be aware that its stability under load needs to be good to keep the signal from chirping, as the BFO and HFO do not like an unsteady DC source. Frequency stability from these circuits is crucial to signal quality. KEYING The first thing I noticed when I received my HW-9, which I did not build, was that the keying was pretty soft. This was especially true when looking at the trailing edge of the keyed waveform on an oscilloscope. There was just too much of a "tail" on the key-up portion of the envelope. Although the HW-8 Handbook has some ideas on correcting this problem, I chose my own method, and it is simpler. It seems that capacitor C578, a 47uf electrolytic, is way too large in value. I found I never needed to turn the Mute Delay control anywhere near its maximum, and so I began to examine this part of the schematic for ideas. By simply changing C578 from 47 uf to 10uf, the key-up waveform edge became much shorter. After this change, I still have plenty of Mute Delay for all practical purposes, but I noticed the shortest setting of the control permitted a little T/R click to pass into the audio chain. It seems the large 47uf original value for C578 delays the turn-off characteristics of the keyed line. By changing R444 from 180 ohms to 1500 ohms, the proper range of Mute Delay is preserved, and the T/R click is eliminated. There is a click at full audio gain, but this seems unrelated to the Mute Delay time constant. It may be due to the DC coupling from U306 to Q303, but I don't think it is significant unless you are hard of hearing and like to run your audio gain full tilt! I also observed the leading edge of the keyed waveform is a little fast on rise time, and sounds kind of "hard" when monitored on a local receiver. A quick improvement for this behavior can be had by changing C435 from 2.2 uf to 4.7 uf. The rising edge of the keyed RF waveform is a little slower now and sounds better. Try several sizes of caps if you want to see what affect it will have. RECEIVE SENSITIVITY Heath has always tended to use the MPF105 junction FET in a lot of their kits. This device has a very poor transconductance, as mentioned in an article I read in 73 Magazine on "How to Make a HW-8 Come Alive", May 1996, by Gerald F. Gronson K8MKB. As he said, Heath couldn't have chosen a worse device for the RF amplifier in the HW-8. "A coupling capacitor would be a better choice than a MPF105", he writes. This is also true for the IF amplifier preceding the crystal filter in the HW-9. Again, using Molex socket pins, I tried a number of FETs to see what could be achieved. Many higher gain FETs, such as the J310 and J309 seemed to have too much gain, and caused oscillation and instability without redesigning the entire amplifier stage and it's biasing, termination, etc. So I settled on a MPF102 (R/S pn. 276-2062, $0.99) chosen from a number of devices I had on hand. A good increase in sensitivity is the result, and I used the current measurement through source resistor R305 as a guide to selecting a proper candidate. I also monitored a weak signal on the rig's S-meter for best signal strength as I chose the best device. Re-adjust T301 and the AGC threshold control, R329 after making this substitution as described in the manual for proper S-meter behavior. Use care and a proper tool to tune the slug in T301 so it does not crack or crumble. AUDIO IMPROVEMENTS The HW-8 Handbook mentions that capacitor C336, a 2.2 uf electrolytic, is installed backwards for polarity. This is certainly true, and it should be reversed from what is shown in the manual, schematic, and PCB silk-screening. If you like, you can try a slightly larger cap for C336 (using Molex pin sockets!) to see if the audio quality is improved. I kept the original cap but turned it around. A mention is made in the HW-8 Handbook about substituting a TL084C quad FET op amp for the LM324 used at U304. This is supposed to retain the high impedance of the active audio filter and is claimed to be a quieter device. I had some TL084C ICs, so what the heck? I did not see a noticeable improvement, but if your audio filter seems noisy or if the bandwidth is not what you think it should be, give it a try. I noticed the active audio bandpass filter center frequency is not the same as the crystal filter's center frequency. I still have to address this problem, but if your rig seems particularly poor in this respect, check R354 and R359 to see if they are close to 1.5 megohms and matched in value. C339, C341, C344, and C345 all need to be matched in value for best filter performance. Heath used ordinary ceramic disk caps for these, and they have only 20% tolerance or so. Use a capacitor meter if you have one, and select four caps that are as close as possible to use in the filter. If you don't have enough of these, remove several other .001uf disk caps from other places on the T/R board and select the four best candidates. Use the fall-outs for the other caps, since their circuits are not as critical. Polyester or mylar caps have normally better tolerance, stability, and leakage than the ceramic disks used, so if you can get them - use them. I think the audio filter's center bandpass frequency should be shifted slightly lower than it currently is, so .0012 or .0015 uf caps might be a better choice here. I prefer not to parallel or series connect several components to get the value needed, so some parts searching is in order. Using 1.8 megohm resistors for R354 and R359 might do the trick, but I haven't done the math or tried them since I didn't have any in my parts bins. Most other audio filters I have encountered do not use such high R and small C values. All new values for the filter might be the way to go, but the existing filter works pretty well - even if it is too high in center frequency for the crystal IF filter and my preferences. The S-meter and audio filter should peak on at the same frequency. TRANSMIT STABILITY and POWER OUTPUT I had no complaint about the output power level of nearly 7 watts maximum on bands 80m through 15m, but 15m seemed a little unstable. On 12m and 10m, the output was down slightly to 3 watts or so. On 15m, the CW Level control seemed to be non-linear when increasing and particularly *decreasing* the power level. I suspected some spurious behavior of the driver or PA stages. It turned out to be in the pre-driver stages. In the original design, Heath used MPS6521 transitors (Heath pn. 417-172) for Q401 and Q402, the pre-drivers that precede Q404. Q404 is a 2N3866 (Heath pn. 417-205) that is more than adequate in power gain and frequency. However, there are ferrite beads on the base leads of Q401 & Q402. I have found this to usually indicate a problem area that needs addressing, rather than taking a Band-Aid (tm) approach! Sure enough, here is where the instability and loss of power output on 12m and 10m was found. If your rig's output level drops suddenly from a constant level as you reduce the CW Level control - especially on 15m for some reason - then you need to look into this. After installing Molex pin sockets on Q401 & Q402, I began the tedious search for transistors that were both more stable and had greater output on the two higher bands. After trying perhaps 100 devices of maybe 5 or 6 types, I finally found the right combination. Q401 seems to be not as critical as Q402 is, and almost any PN2222, 2N2222a or MPS2222a will work fine for Q401. However, Q402 is a different subject entirely. Nearly every flavor of xx2222x device I tried caused spurious output and instability - especially on 10m. The ferrite beads offered no help in reducing this problem. I finally found (after some mumbling and cursing) 1 or 2 2N3904 transistors that behaved well and gave decent output on 10m. Many of the 2N3904's I tried were simply "too hot" and caused a spur on the transmitted waveform between 3 and 5 watts output. The OHR WM-1 wattmeter also indicated a sharp increase in output level, due to increased harmonic / spurious output. The extra "trash" on the transmitted waveform was easily noticeable. Apparently, without major changes in the circuit design, about 4 or 5 watts on 10m is the greatest output level that can be produced without excessive spurious content. I settled for this, using a prime MPS2222a (R/S pn. 276-2009, $0.59 ea) for Q401 and a MPS3904 (R/S pn 276-2016, $0.59 ea) for Q402. If you have several different manufacturing brands for the 2N3904 device, try them. Some behave quite differently from others, and you need to find one that is not too "hot" but has decent high frequency performance. Expect to spend some time trying quite a few before you find the *correct* device, and the Molex pins or sockets are a must for this selection process. Note that the ferrite beads were of no help in reducing the spurs, and are *not* required or recommended with the transistors I recommend. The 2N4401 part mentioned by some was a poor performer compared to the 2N2222a / MPS2222a and 2N3904 / MPS3904 combination for Q401 and Q402, respectively. My final choices were *both* made by Motorola: Q401 is a MPS2222a and Q402 is a 2N3904, although Radio Shack lists it as a MPS device, it is actually branded as a 2N3904 device. The final result is a broad range of adjusting the CW Level Control will produce from 7-8 watts on 80m through 12m, and about 4.5 to 5 watts on 10m, depending on the VFO setting. No instability or sudden changes in output level occur on any band, as it should be. As mentioned in the HW-8 Handbook , I use and recommend NTE401 heat sinks for long life of the MRF237 PA transistors, especially if the ambient temperature is high (as in AZ on BUBBA day! ) or if you enjoy long tune-ups, 5 watt operation, or rag-chewing. They will just fit if you adjust the position of T403 slightly. Use a small amount of themal compound before pressing the heat sinks into place. The bandswitch shaft will just clear the new heat sinks, that have one more fin and are slightly larger in diameter. Use one of the original heat sinks on Q404 if you like (it's not really needed) and rest confidently that the PA transistors Q405 & Q406 are much cooler in operation. Be safe and hold the output level at 5 watts maximum. I'm curious as to why Heath didn't design the PA stage as a push-pull output, as it is almost as simple as the parallel PA design they used. One day I might try to change it to push-pull as that would insure the two transistors share equally in the load, and second-order harmonics would be even less. Some biasing changes might be necessary, but the parallel scheme works just fine, although if the two MRF237's aren't pretty close in characteristics, one of them will "hog" more of the load than the other. If your rig's output is low on all bands, one MRF237 is probably toasted. END OF PART 1 AB7MY Gary Surrency Chandler, AZ (Near Phoenix), QRP-L #571, AZ ScQRPions, ARRL VE