Tech Tips on the SX Series
by Tom Osenkowsky, Mike Patton, and others
(If you have additional tips that you would like to share on the SX series, or any other transmitter, please send them to us.
The following is a troubleshooting and maintenance guide for the Harris SX series of AM transmitters.
I'll start with a tip for proper Loading control adjustment. Especially in higher frequency SX transmitters there may be two different locations on the Loading control where minimum reflected power can be obtained. The correct location for the Loading control is where you experience an increase in current with lessening inductance or turning the Loading control to your right produces an increase in current.
Here is a listing of some problems and cures based on experiences:
Problem: Transmitter randomly turns itself on or off/False PDM Kill.
Cure: A noisy optoisolator on S&M board is the cause. Replace appropriate chip. Check bypass capacitor on ground side of “Off” remote control terminal on Customer Interface Board. A dirty carbon pot on Phase Monitor module can cause intermittent shut downs. A false PDM kill can also be caused by intermittent Aux contacts on AC contactor.
Problem: Unable to clear overloads.
Cure: Check/replace U45 and U46 on S&M board.
Problem: High power interlock appears to be engaged when it is not. Transmitter
acts weird. Transmitter will not engage in high power mode. Low & med power
lights on, red Fault light on.
Cure: Replace U42 on S&M board. Correct logic state will appear on U42 but chip is bad. Can swap with U46 but you will not have any Overload protection. These chips are not in spare parts kit.
Problem: TX inoperative. No 5 volt supply.
Cure: Check 5V fuse on AC controller board. Fuse is in 6V switching supply. Also, the 7815 and 7915 regulators have been known to fail.
Problem: No PA drive.
Cure: Check small .1 uf bypass caps for shorts on RF Oscillator board. Measure R8 and R23 for correct value. D4 can short if resistors drift low in value. Caps around regulator can go bad especially electrolytics which can cause a ragged waveform and low output drive from oscillator. RF output should be square, not ragged. Replace C17 and C18 if drive levels are low. Use 105 degree caps. Original DS0026 can be directly replaced with On Semi MC33151, Texas Instruments TC426 or Intersil EL7202. Also check Q11 and Q13 on IPA board. Ensure Int/Ext jumper on Oscillator board is in correct position.
Problem: No PA Drive.
Cure: from Ron Huckaby, of Butte, Montana. He worked on this obscure problem, and it turned out to the 8-fold coax from the IPA power divider to the PA input. Cable tests 100% good, but will not drive the PA. New cable, testing equally as good, solved the problem. The solution came from one of the old SX-1 design engineers who knew of this quirk.
Problem: Main AC contactor chatters, tx will not engage, erratic symptoms,
display readings. Symptoms may disappear when top Controller board is
disconnecting by removing top blue connector.
Cure: Replace the MSW-5101 U12 chip. Socket this chip. Intermittent thin PC board solder traces and connections have also been known to cause erratic, intermittent operation. Especially check the solder jumpers E-1 to E-4. Also check battery for proper storage voltage. If batteries are low tx will turn on to lowest power setting on whatever power was selected. Ensure battery charging resistor has been soldered to pad when tx was installed.
Problem: TX will not turn on.
Cure: Check auxiliary L4/T4 contacts on K1 and K2 AC contactors. Ground switch logic may be affected by oxidized contacts providing false information to controller.
Problem: Out of band emissions. Erratic problems with RF Oscillator.
Cure: CR1 on RF Oscillator board can oscillate around 3 kc. Newer SX tx's have a 5 watt CR1. Be sure drive pot on RF Oscillator board is high enough so that square wave becomes less like a sine wave. Freq can double causing major drive problems if drive is too low.
Problem: PDM amplifiers fail.
Cure: Ensure PDM amps have correct size fuses installed. SX-5's have a 4 amp AB fuse, not a 2 amp AB. Sometimes engineers confuse the 2 amp AB used on the RF Amp boards with the 4 amp AB used on the PDM amps. Ensure tight ground lug connections on all the nearby 1500 uf caps. The U11 chip on the PDM amp can sometimes oscillate or generate noise on the PDM signal. U11 is *NOT* included in the Harris spare parts kit. The diodes on the PDM Pullup board can fail causing PDM amp transistors to fail. Arcing underneath the PDM amp heatsink can cause failure. PDM amp damper diodes can fail. Older diodes are no longer available, Harris makes a retro kit to fit newer style diodes as replacements. Diode CR1 is critical to module protection, check and replace if necessary.
Problem: Multimeter #26 reads appx 234, PA volts very low, PA amps high, power
output about 300 watts on SX-1.
Cure: CR13 and CR19 shorted on PA RF Output board.
Problem: Output flicks on/off with no faults indicated. Problem clears after
doing a #99 reset. Problem can also be cleared by inducing a fault.
Cure: Older SX tx's R211B and R221B on the RFI board should be 910 ohms, not 510 ohms. R212 and R222 should be 51 ohms, not 510 ohms. Older schematics may not reflect these changes.
Problem: Display garbled, with the exception of #40-44.
Cure: A/D converter chip bad. Replace.
Problem: Pinned PA voltmeter. PA fuse finder reads appx 2.
Cure: Shorted diode(s) CR41 and/or CR21. This will also shows as bad MOSFET’s when ohmmeter tested. Two PDM amps read zero on multimeter. Two phase volts read appx 260. Problem caused by burned up PA output network capacitors. When this happens, numerous PA MOSFET’s and fuses will have to be replaced.
Problem: Upon turn on, tx produces fault #2. PA Volts & Amps pulsate in fault
condition. Can reduce power on Low or Med position. Power varies rapidly and
does not correspond to reading #44.
Cure: PDM Generator board pot R28 is bad.
To be certain L1-C1 output network components are properly set, be sure PA tuning somewhat coincides with peak in PA current.
Cure: The back-to-back Zener diodes that clamp ringing on the small RF transformers usually OPEN in failure mode. This is contradictory to the usual Zener failure mode of shorting. Never operates tx with bad PA MOSFETS removed from socket. Leave shorted MOSFETS in socket. Diodes on PA modules may tend to fail. Check diodes whenever MOSFET’s fail. Also check for shorted transformers. Very early production runs had turns twisted too tightly causing premature failures. Be sure E-2 on PDM pull-up board is mechanically secure or else damper diode may function intermittently or not at all. When troubleshooting system controller boards, open a door to prevent tx turn on at power to prevent damage. If any output network components are replaced, it is necessary to check tuning using a Vector Impedance Meter. MOSFET’s can fail by excessive low audio frequency modulation. This is usually the first symptom of failing power supply capacitors. PA fuses can fail from bad p/s capacitors even if they are not sloshing.
Problem: Transmitter won't come on, shows Fault 2 PA Current analog meter is
Cure: The Analog Devices module U3 on the AC Controller board can become invaded by moisture and cause the voltage output to be inverted to negative values. This module is soldered in place, so do not try to pry it off. Use a hair dryer on warm setting to heat the module and drive out the moisture. (Steve Brown)
Harris SX2.5. Various overloads inhibit startup. Tries to come up, fault light
with overload indication immediately thereafter. No evidence of a real problem.
Cure: Check the resistor network R37 that serves as pullups for U44 and U45 on the S&M board. Without a pullup, (in this case, fault 4, underdrive), output will not go high and U46 will see it as an overload. If a replacement network is not available, 1k 1/2 watt (2% in the original circuit) can be tacked across the back of the S&M board of the offender from +5 to the output in question but recommend bypassing with a .01 ceramic. (Len Watson)
Problem: “Off” Light flashes, PA Volts & Amps Meters Pulse on and off, tx shuts down with Error 64.
Cure: Meter control chip on the big control card was blown apart.
It's a small 8 pin chip in the meter circuit. Close to the bottom of the card. Replace it.
Problem: No RF Output. Cure: Dirty ribbon connection causing RF mute, no 2 volts at Oscillator Board R16.
On an SX-1A problem was no RF output, no PA Current reading. Problem was a bad PA current metering module. Module was a
Harris lab grade current sampler. It is located near the bottom on the right side of the transmitter if my memory is any good. It has an internal shunt that was open.
Problem: Front Panel PA Voltmeter frequent failures. Cure: Check the connections on the interface board located on the demising wall between the pull out drawer and the front cabinet. There is a white barrier strip type Molex plug located on the lower front of the board which has been known to go intermittent. The PA Voltmeter meter sample passes through that barrier strip.
Problem: False Mod Fault light. Cure: A number of RF FETs not
conducting will show high volts/low current. Replacing the RF FETs would clear the mod faults. If you turn the PDM off all the way the current should be zero. If it is not you have a shorted MOD FET.
Problem: SX/Gates 5 burning odor. Cure: Problem was poor connection on banana plug/jack on PA board. Check high current point in Gates where PDM pullup boards meet large filter caps. One end welded itself back together, other end failed. There is one of those "Buss bar" fuses on the Right side wall about 1 foot up from the floor that blew. It was in the ground return for the PA section. Failure was caused by a B+ wire laying against it and slowly melting into it over the years. Since finding that I've check a couple more and found the same problem that hadn't turned into a boom yet.
Don't forget to tighten RF buss at top or bottom of PA boards in SX-x/Gates transmitters.
Problem: SX/Gates PA board toroid cracked when tx room was very cold. Cure: Heat room to reasonable temperature. Blower bearings can scream during very cold temperatures.
Problem: SX-2.5 low RF output, output coil L2 runs hot.
Cure: Oscillator board generating excessive harmonics. Replaced board, tx operates normally.
General Notes: An intermittent Bessel Filter switch on PDM Generator board can cause strange symptoms. Replace switch with a soldered connection in Flat mode. Strange symptoms can also be caused by open filter capacitors in 30V power supply.
Problem: Trying to "read" the Phase 2 PA Volts on an SX-5. Mike Patton Offers this overview:
OK, folks, here's how it works on the SX series. I'm going to give a bit of
background here, so please don't be offended if I mention things that
most of you already know.
The B+ is about 250 volts. It is connected to the big banana plug on each PA board, fed thru fuses located on the floor of the cabinet. (IMHO, in the SX series, there are entirely too many fuses, which Harris changed in the Gates series, which does not have these fuses or the 10A fuses between the PDM LP filters and the PDM amp outputs.) The PA voltage for full power is about 100V on all models in this series and in the Gates series, too. PA current varies with the power, about 12A for full rated power (1100 W) in a -1, 30A for 2.75 KW in a -2.5, and 60A for 5.5 KW in a -5.
All SX transmitters generate four "phases" of PDM drive. When we use the word phase, here's what we mean: One of the problems with switching
modulators is that there is always some unwanted residual ripple at the PDM switching "carrier" frequency that leaks through the PDM LP filter and modulates the PA just like the wanted audio. This ripple will generate unwanted/illegal sidebands at +/- the PDM switching frequency (usually 60-80 KHz) from the RF carrier.
With a conventional PDM circuit, the combination of the PDM LP filter and the RF output filter (usually a bandpass filter in most PDM TXs) had to attenuate these sidebands below the FCC requirement limit. This forced the designer to have to compromise the HF audio performance of their designs to get the needed PDM sideband suppression.
Harris, the originator of the PDM modulator (at least as I've been told),
came up with a nice innovation in the SX series: multi-phase PDM drive. It
works something like how a 3-phase DC power supply has a higher ripple frequency
and can therefore get by with a smaller filter cap than a single-phase supply.
Here's how: the instantaneous value of the incoming audio changes the duty
cycle of the switching modulator, which changes as the audio level varies;
that's how any switching modulator works. The switching signals for each phase
all have the same duty cycle, representing the same audio, but they are
generated using four different 60 KHz PDM switching carriers that are 90?
apart. The modulated RF driven from these four
phases is recombined by the fact that the outputs of all the PA FET quads are combined in series on each PA toroid board (and on -2.5 & -5 models, all the PA board outputs are also in series.) This produces a modulated RF signal whose apparent PDM sideband spurs are 240 KHz from
the carrier, instead of 60 KHz, just like a 3-phase supply's ripple freq is 360 Hz instead of 120 Hz for a 1-phase. The output network on this TX, as are those of many modern SS AM TXs, with an RF passband narrow enough to attenuate these sidebands pretty well. This allows the use of more gentle slopes on both the PDM LP filter and the RF output bandpass filter than would otherwise be required to attenuate a 60 KHz PDM switching freq, which gives better audio performance (less HF distortion, less ringing (better mod peak control)) and stresses all the filters less.
All SX transmitters need and use all four phases of PDM drive, but how they are generated and fed to the PAs are different in the different boxes.
As in most if not all PDM units, the PAs, the PDM LP filters, and the switching FETs on the PDM amps are all wired in series. In the SX series and their children the Gates series, the + side of the PA board(s) are wired from the B+ supply (big banana plug) "down" to the four phase inputs (small banana plugs), which are connected to the output of the PDM LP filters, whose inputs are fed from the PDM amp FETs, which are grounded. The PDM amps always switch from ground to B+ at a 60 KHz rate, and the LP filters strip out this switching carrier, leaving only the audio to modulate the PA (with the issues noted above).
Each PDM amp board contains two separate amps that can be driven from one
signal or split into two separately-fed units--there is a little red
jumper that selects this. Each PDM amp section has two FETs in parallel. Originally, these FETs were just hardwired together, so if one failed, the pair was shorted to ground, and (as has been noted by others) either the fuse on the PDM pullup board would blow, or the PA fuses for the affected pairs would blow--or occasionally the fuses will hold and the FET pair would run flat-out until the problem got fixed, which makes only unmodulated RF (and lots of heat)--not good. On later versions, each PDM amp FET is separately fused; the idea is that, should one FET fail, its fuse will blow and leave the other, unshorted FET, to carry the load, which it can do, sometimes indefinitely, with no deterioration of performance. Good idea, that.
On an SX-1 and the SX-2.5, there are two PDM amp boards, with both sections on each board setup as separate amps, to make all four phases. On the -1, the two PDM amps' four outputs are fed through the PDM LP filter board (one filter board contains four separate filters), and the filters' outputs are fed to the four phase inputs on the one PA board. On the -2.5, there are two PA boards, and each LP filter's output feeds two loads: one phase on each PA. On the -5, there are four PDM amps, each carrying one phase (in the -5, the jumper is set to have both amps on each board drive from one input), and two filter boards, giving a total of 8 separate PDM amps and LP filters, with each phase being generated and filtered twice. This is because each PDM amp/filter combo can drive only two PA phase loads, but you've got 16 loads total (4 on each PA board x 4 PAs). With 8 sources of PDM drive, each filter's output again feeds two loads, one phase on each of two PA boards.
The SX units have lots of good internal metering, at least when the digital display is working. The PA voltage indications are obtained for each phase by op-amps located on the AC controller board and setup as bridges, which compare the B+ voltage to each phase's DC voltage, sampled at the LP filter's outputs, thus obtaining a DC sample output whose level represents the difference between the B+ and the four phases fed to the PA boards--i.e., the PA voltage. On all models, these are indicated by positions 06-09 on the digital meter, showing phase 1-4, respectively (IIRC). The front panel analog PA voltage meter is a repeat of the digital meter for only one phase--phase 1 (again, IIRC).
BTW, you can have a good PA voltage analog meter reading while still having
the other phases be bad--or vice versa, which is why one always
should check all the phases on the digital meter to know the whole story--and even that isn't enough for a -5--see below. The outputs of the switching modulators are also metered, indicated on positions 26-29 on a -1 and a -2.5. In a -5, 26-29 show one of the two outputs from each PDM amp (Q19 & 20), in order (A6-A9), and positions 36-39 show the other set of PDM amps on each PDM amp board (Q39-40). When the
switching modulators are running right, these readings will be about "16" on each of these positions. If any is 0, that's probably a shorted modulator. A very high reading probably indicates a blown PDM fuse.
Here's the non-obvious problem: on a -5, only one "set" of PA voltage phases (remember, there are two sets, to carry the multiple loads) is metered by either the digital or the analog meters. The other set, feeding two of the four PA boards--that is, fully half the generated PA voltage sources--are not metered as such, although their level can be inferred by the PDM amp output samples, noted above, but this is not always reliable. The only way to be sure that all the modulator's outputs are right on a -5 is to open the doors while the unit is running and measure the voltages.
If you've got a problem, first check all the fuses--and be sure they are the right types. Note: on the PAs, it is important to use the extremely-fast-blow "rectifier" type fuses (Buss GBB series, Newark has them for a pretty good price) recommended by Harris. By the way, these extra-fast fuses are NOT there to protect the PA FETs from failing--a FET can and will fail if stressed much faster than ANY fuse can open. They are there to (hopefully) prevent cascading failures of the PDM FETs once the PA FETs short (and they always short), or even worse, to prevent physical damage to the PA board traces--that monster PS can send 300A+ into a short--well enough to smoke a hole in a PA board; I can post pix! Also note that, some years back, Harris/Gates updated the recommended rating on these fuses to 4A from the old 3A rating. There are also fuses on the PDM amps' pull-up boards. These fuses are 2A for the -1 and 5A on the higher-powered types. Although Harris recommends rectifier fuses here, too, it's been my experience that, since the PAs and the PDM amps are effectively in series, one can use with impunity standard AGC types here--much cheaper.
If you have any blown fuses, check the FETs for shorts, which can be done with the PA and PDMs mounted--the procedures are in the TX book.
Replace any that are dead, obviously. On an unfamiliar box or one that's really unhappy, I tend to ohm out each modulator and PA FET pair, just to rule out any problems before I put the juice to it.
Once you are reasonably assured that all is good, the best (only) way, in my experience, to truly verify the proper operation of the PDM circuits in one of these rigs is to carefully defeat the interlocks, and run the unit with the rear door--or both doors, for a -2.5 or a -5--open. BE VERY CAREFUL: the 250VDC B+ is widely exposed in these boxes, and you can as easily be injured or even killed by this voltage as by a higher HV. You can use a scope to look at the switching waveform at each PDM amp's output, and a scope or a DVM to measure the DC voltage at each phase input jack on all the PAs--they should all be the same. Measure the B+, then the voltage at the phase input; the PA voltage is the difference. Here's a trick: Many meters have a "relative" mode, which allows you to set a "fake" zero while measuring a voltage, you can then use this artificial zero to compare other voltages to your reference without having to connect the ground lead to a high voltage or doing math in your head. FYI, the modulator will have a nice pretty square wave output, but only if it is fully loaded. If you find a PDM output with an ugly waveform, look for a blown PA fuse in its load circuits.