Send DX Hints or Kinks to any of the Blog Authors shown below.

Friday, January 20, 2017

Repairing an Alpha 87A for a Friend

After repairing an Alpha 87A for a friend, I thought memorializing my notes would be prudent, for future reference.  -  Charlie, N8RR
These are very nice amplifiers.  The silent QSK and T/R switching is impressive. The power output from a pair of 3CX800 tubes is robust. Band change and autotuning is extremely fast.
Alpha 87A Autotune HF Amplifier
The Alpha 87A is capable of 1.5kW continuous RF output on all commonly used modes and on any authorized amateur frequency from 1.8 to 29.7 MHz. Band change and tune-up are microprocessor controlled. Capable of full CW break-in (QSK) and all digital modes when used with any appropriate transceiver.
The amplifier has been around a long time. The amp which I recently repaired in Dec16/Jan17 was built in early 1994. The first Alpha 87A amps were built in October 1991.  The tubes are still available new, but are expensive. The amp was designed to use Eimac tubes. It is reported that less costly Chinese 3CX800 tubes do not work well in the 87A but I don't know if this is true.
Alpha has recently been through ownership changes. Factory support for this vintage amp may become problematic. Others, such as Array Solutions and Brad Focken, KØHM,, will support the 87A.  Brad is one of the designers for the Alpha 9500.
Parts availability is an issue.  Even if qualified service support is available, some specialized parts could be a problem.  Still, many of the parts remain available and there are modern replacements for some of the obsolete parts.
My experience with this amplifier and dealing with Alpha factory support is limited. This memo summarizes what I was able to learn in the process of repairing this Alpha 87A.

The Alpha 87A is much more complex than a typical manual tune amplifier. If you have a problem with yours, and you don't have any experience working on high power tube type amplifiers, the 87A is probably not the best candidate to start learning on. Still, there is a common problem with the 87A that most folks can address.
Wiring Harness Routing Behind AC Module (Removed)
The various boards are interconnected with a wiring harness using Molex plugs and terminals. Over time, these plug connections can oxidize and become intermittent, causing all kinds of operational issues. These connections can be restored by unplugging the connector and applying a small amount of a good electronic contact cleaner, such as Deoxit, to the terminal pins.  Sometimes just unplugging and reinstalling the plug will restore the electrical contact.
Before attempting this or any more advanced repair, proper safety precautions must be taken to avoid injury or fatality. Before removing the cover, unplug the amplifier from the power source. With the cover removed, a safety crowbar shorts the high voltage to ground, making sure the power supply capacitors are fully discharged. I always short the high voltage to ground as a final precaution. The cleaning/reseating of the Molex plug connections can safely be done at this point. No Power On testing with the cover off is recommended if you have no experience working with HV.
If you are comfortable and experienced doing power on testing, it is necessary to defeat two safety interlocks in order to power the amplifier with the cover removed. One is a microswitch on the AC input; the other is a HV crowbar that shorts HV to ground. The plastic cap for a pen can be inserted to block the crowbar switch open. Note there will be HV present on the plate RF choke, tube plate circuit and the HV rectifier board at all times when the amplifier is turned on, even in standby mode. It is strongly suggested those with no HV or amplifier experience get someone with such experience to mentor and help if it is necessary to troubleshoot the amplifier in an energized state.

This is worth restating. If you have no experience with tube type amplifiers, get some qualified help before considering any power on testing with the cover removed or find a qualified repair vendor.

For those with manual tune amplifier experience, there are many 87A problems that can potentially be found and addressed.

There were no problems with the autotune circuits with this Alpha. Thus, I did not need to dive into this aspect of the operation.
If the amplifier tuning is on a different band segment from the current RF drive frequency, this condition is instantly detected and closure of the input/output relays in the OPERATE mode is inhibited until the amplifier changes bands and retunes. On CW, one “dit” from the transmitter is enough to cause band change and retuning. On the second “dit” transmitted, the amp will be in OPERATE mode and making power. This tuning feature keeps the amplifier from making power into a mismatched load.
Once the amp has reset to the correct band and tuning, the input/output relays remain continuously closed as long as the OPERATE mode is maintained. There is a front panel LED which indicates when the amp is in OPERATE or STANDBY mode.
More Wiring Harness Routing
The amplifier has an input and an output wattmeter. These wattmeters remain in the circuit in both Standby/Bypass and OPERATE mode. The RF sensing on the input is tapped off of the input wattmeter board, so the frequency counter and associated logic is active when the amp is turned on.
Note that if the input and output relays fail to close in OPERATE mode due to a malfunction, the OPERATE LED can still light and the internal T/R switching will still function, but the amp main RF path remains in bypass mode. If the input sensor sees RF drive in this condition, the amplifier will immediately fault to STANDBY mode.

Other than the autotune and associated protective features, the main difference between the 87A and most other amplifiers is the use of PIN diodes for T/R switching in place of relays. Alpha has also used PIN diode switching in some other of the modern manual tune amplifier designs. Ameritron also uses PIN diode switching in their QSK5 accessory which can be retrofit to their (or other) amplifiers. Alpha reverted to conventional relay T/R switching in the 9500 Model.
This was my first time working with PIN diode T/R switching.  Although the problems with this 87A were not related to PIN diodes, I didn't know that at the beginning so some time was spent studying PIN diodes and figuring out how the system was supposed to work.
The PIN diode looks and works like a regular diode. A diode passes current flow in the forward direction and blocks current flow in the reverse direction. When an AC signal is applied to a diode, the diode will conduct on the forward half of the cycle, but the current flow will be blocked on the reverse polarity half of the cycle. This is how a rectifier works, turning AC current flow into pulsating DC. For the diode to do this, it has to have a fast recovery time when switching from conducting to blocking mode.
A PIN diode works the same, except the recovery time is deliberately slowed down. You don't want the diode cutting off part of the RF cycle when the polarity of the signal changes. You want the diode, once biased to conduction, to stay in conduction long enough so that the full RF cycle will pass through and not be cut off on part of the cycle. A PIN diode is just a slower switching version of a regular diode. This is a grossly simplified explanation.
The PIN diode is an on/off switch for the RF path. To turn the diode on and off, a DC bias current is used. When a sufficiently high negative DC voltage is applied to the PIN diode cathode relative to the anode, the diode is forward biased and will conduct.
Present in the PIN diode circuit are both a DC current for control purposes, and the RF signal. RF and DC voltages and current are simultaneously present in electronic circuits. An example is the tube plate circuit in an amplifier. There is the RF output and the DC from the HV supply simultaneously present. You don't want high voltage DC on your output tank components or RF output connector, so a plate blocking capacitor passes the RF out to the tuned circuit but blocks the DC current. At the same time, you don't want RF getting into the HV power supply, so a plate RF choke passes the DC current for the tube plate, but blocks the RF from getting to the power supply. RF chokes and capacitors are used to block or pass RF or DC, as required. Another common application is passing 12VDC from the shack to an external preamp or switch device over a coax cable. Same theory, DC and the RF both pass over the coax, capacitors and RF chokes separate the voltage/current at the circuit ends as appropriate.
The Alpha 87A has three different DC bias sources used for controlling the PIN diodes. These are applied during receive or transmit as appropriate. Note that it is not sufficient to simply remove forward bias from the PIN diodes to assure they are cut off. A significant reverse bias voltage is applied to force the diode to cut off.
There are 3 PIN diodes that are turned on (biased on) in transmit mode. Diode D1 is on the Input T/R board and looks physically like a conventional diode. During transmit, it is biased on to pass RF from the input circuit to the tube cathode, for driving the tube. During receive, this diode is turned off by reverse bias so that RF from the common RX/TX RF path cannot make it to the tube.
Over in the Output T/R board enclosure, there are two larger, stud mounted PIN diodes that are turned on in transmit and off during receive. During transmit, these carry the RF output from the amplifier tuned circuits to the output wattmeter/RF out connector. During receive, these are biased off to isolate the amplifier output from the antenna circuit.
Also located on the Output T/R board are PIN diodes D2, D6, D8 and D11 which are forward biased on during receive mode and reverse biased off during transmit mode. These are the same component as the input PIN diode in the tube cathode circuit. When the amplifier is in the OPERATE mode but not transmitting, the receive path between antenna and radio is through these 4 PIN diodes.

There is a ‑109 volt bias source which is used during receive mode to forcefully reverse bias the input PIN diode in the tube cathode circuit and also to reverse bias the two large TX PINs over in the output T/R enclosure. The ‑109V is applied to the anode of the diodes to accomplish the reverse biasing and thus cut off. In transmit mode, the ‑109 volts is sunk to ground potential by transistor Q2 on the input T/R board, removing the reverse bias. The actual voltage in this amplifier was nominal ‑104 volts during receive state. References showed the range of this voltage could be down to ‑89 or ‑90V and still be OK. A good place to measure this voltage is over on the safety RF choke on the output of the amplifier.
Another bias source, approximately 34 VDC, is active on both transmit and receive modes, but the polarity of the voltage reverses between modes. This is called Tbias.
In receive mode, Tbias is positive. Positive Tbias is applied to the cathode of the large TX diodes, which along with the ‑109 volts already applied to the anode of these diodes increases the total reverse bias to ‑143 volts, combined from two sources.
Also in receive mode, Tbias applies +34 volts to the ANODES of the four RX PIN diodes, thus forcing them on for the receive RF.
Output T/R Module Showing RF Choke and Spark Gap Device
Continuing in receive mode, Tbias also applies +34 volts to the cathode of the D1 input PIN diode over in the tube cathode circuit. The positive voltage on the cathode adds to the ‑109 voltage on the anode of this diode, increasing the total reverse bias to about ‑143 volts on the input PIN diode during receive mode.
Tbias in transmit mode reverses polarity. Bias on the input diode becomes ‑34V on the cathode, thus forward biasing the diode and sending RF to the cathode circuit.
The ‑34V Tbias also drives transistor Q2 into conduction on the Input T/R board, thus sinking the ‑109V bias source to ground, removing this big reverse bias from both the input diode and the two big TX output diodes.
The ‑34V Tbias also is applied to the cathode of the two large TX diodes, forward biasing and turning them ON, thus completing an RF path from the amp output to the output wattmeter/output connector.
Finally, ‑34V from Tbias also is applied to the anodes of the four receive PIN diodes, thus cutting them off.
Last but not least, we have the Rbias source. This is a very large positive voltage that is applied to the cathodes of the four receive PIN diodes during transmit mode. It is on the order of 870 volts in this Alpha 87A and is used to forcefully reverse bias the receive PINs so there is no chance the amplifier RF out can get back to the amp input. Actually, I think this source is read as Not Receive bias.  

‑109V bias source: Present during receive to reverse bias the big TX diodes and the input PIN diode. Off during transmit.
+/- 34 V Tbias source: Positive during receive on the cathodes of the big TX PINs, on the anodes of the receive PINs, and on the cathode of the input PIN diode.
Negative during transmit on the cathode of the big TX PINs, on the anodes of the receive PINs, and on the cathode of the input PIN.
Rbias source: 870V Positive on the cathodes of the receive PINs during transmit. Absent during receive.
Remember, on a PIN diode, cathode negative relative to the anode = forward bias and switch is on. Cathode positive relative to the anode= reverse bias and switch is off.

The 87A CPU receives inputs from the various bias sources and makes sure the levels are within safe operating parameters. If it senses a low level or missing bias source, the amp will fault. In some cases it will not allow Operating mode status if a bias voltage is not correct. For example, the Rbias monitor wire was broken on this amplifier. The amp generated a hard Fault 1 when the OPERATE switch was pushed.
The amp monitors HV level, plate current, the various bias voltages, RX PIN diode state, filament current, gain/tuning, reflected power, temperature, etc. This list is not all inclusive. The amp should be relatively bullet proof with all these protections.
You can measure many of these monitored parameters at the J1 and J2 connectors on the CPU board with the top cover removed from the amplifier. Be careful not to short between the PINs when making these measurements.
First Assistant Technician - Sheba

Regarding testing of the PIN diodes, I became convinced pretty early there were no issues with the PINs. There are conventional resistance tests that can be made, but I did not do those, not wanting to unnecessarily unsolder one end of the diodes from the board. Instead, I measured the voltage across each diode “cathode to anode” to determine the forward bias. There is a series resistor between the bias source and each diode or diode pair. I measured the voltage drop across these resistors and calculated the forward bias current in each diode. There are curves published for each diode that plot the forward bias current versus the forward bias voltage. The values I obtained for forward voltage and current seemed reasonable when looking at these graphs. My thinking was that if the diode was leaky, it would show as an abnormality in my readings. Whether this would turn out to be true in practice all the time, I am not sure. In this case, I did not chase a PIN diode problem because my measurements seemed reasonable, and I got lucky.
For example, on the big TX PIN diodes, I measured .9 v forward bias and 740 ma bias current through each in transmit, using the above method. These values seemed reasonable looking at the published curves.
The PIN diode DC bias voltages should be measured with no RF drive to the amplifier. I used a footswitch to key the amplifier.
Regarding the four receive PIN diodes, I thought the fact the bias voltages looked normal along with the proper functioning of the receiver in OPERATE mode, indicated these were OK. When the amp is in OPERATE mode but not transmitting, the antenna path to the receiver is through the 4 receive PIN diodes.

After finding and fixing the broken wire that caused Fault 1, we encountered soft Fault 17 each time RF was transmitted to the amp after it had automatically changed bands and retuned. First “dit” = band change retune. Second “dit” = instant Fault 17.  No PIN diode faults, etc.
It appeared the amp was in OPERATE mode per the front panel LED and the computer terminal I had communicating with the CPU. However, it turned out the input and output relays were not closing, even though I thought the amp was in OPERATE mode. The bias would change state between receive and transmit. The tube would draw idle current in transmit.
Apparently, there are several issues that can cause Fault 17, which is a high/ low gain or severe mistuning fault. I first got sidetracked for a short time on a discrepancy between the input and output wattmeters, not noticing the relays were staying de‑energized. However, the wattmeter ratio differential was not sufficient to cause this problem. To date, I have been unable to find out the actual factory parameter settings programmed into the logic for allowable wattmeter ratio differential. The cases I have found referenced had a totally malfunctioning wattmeter on the input or output.
The input and output wattmeters can be read using a computer running Hyperlink terminal program for communicating with the Alpha 87A. Various other operating parameters can be read, such as plate current and grid current. In addition, the software allows remote operation of the amp from the computer keyboard. Another feature is the monitoring of the amplifier operating state, and direct readout of the fault code which is causing a shutdown.
Once a sufficient number of hard faults has occurred, the amplifier will generate a Fault 99 and it will no longer power up until the processor is reset using a set of commands which are available from Alpha. Alpha has asked for these factory reset commands not to be published. It is highly recommended for troubleshooting the amplifier to use the Hyperlink program and to obtain the factory reset codes. For this amplifier, during the Fault 1 troubleshooting, I would have been dead in the water without having the ability to reset the processor after a Fault 99. In fact, I had to reset the processor twice.
Mistuning of the amp, or having the tuning parameters out of whack, can cause Fault 17 but fortunately I did not get sidetracked chasing a non‑existent problem with the tuning.
The open relays should have been obvious, but due to my unfamiliarity with the amp, the CPU and front panel LED being in OPERATE mode faked me out. 
Moral: Check the simple stuff first!  
When the amp was driven with RF, the condition was detected and the amp immediately gave a soft Fault 17.
The issue was another broken wire in the harness, between the two relay coils, which are in series. The amp had previously undergone a blower change by the owner, and it is suspected the harness was damaged. When checked twice, no obvious pinching of the harness was found. Both the Output T/R module and the tube deck had to be moved for the blower change, so it could be simply flexing the harness wiring caused the problem.
Sheba Resting After a Hard Alpha Troubleshooting Session
How I discovered the open relays was due to testing the input circuit with a SARK antenna analyzer. I wanted to see if the output level of the SARK was low enough to avoid triggering the RF detection on the Alpha. If it was low enough and the amp would stay in OPERATE without faulting, I should be able to "see" the impedance of the load at the tube cathode during transmit. This would verify the input tuned circuits and double check the input PIN. On receive, the SARK would have a path to the dummy load through the receive PINs, so it should see the dummy load on receive.
When I tested, there was no difference in the SWR/impedance curve between STANDBY or OPERATE, and no difference between transmit and receive. Wait a minute, there is no way these curves should be identical. The RF circuit had to be in Bypass, even in OPERATE. Presto, I immediately found the relays not closing. It should not have taken this test to find the relay problem but it did. Chalk one up to not following my own standard procedures, check the simple stuff first, especially the relays.
The open relay circuit wire was replaced and the amp was back in business.

FINAL NOTES: The actual Tbias voltage on this amp was less than 34 volts. It is a range. I think it should be somewhere between 29 volts and 34 volts.
Likewise, the ‑109 volts bias is a range. This one was ‑104 volts. A reference I found said anything above ‑89 volts is OK.
The KS1001 PIN diodes used in the input and receive switching are obsolete. A reference I found said that 1N4007 diodes will work in place of the KS1001. Alpha apparently has used 1N4007 diodes as replacements. Apparently the switching characteristics are OK as an RF switch, but I don't know if all manufacturers are suitable. If this amp would have had a defective KS1001, I would have tried the 1N4007. Not recommending this and your mileage may vary. I am just repeating information that was found.
The two big TX PINs are obsolete as well (MA4P4006D). I found references stating a UM4006D or UM2106D (Arrow Electronics) is a suitable replacement. Again, I did not need to pursue replacing the TX PIN, so this is information that is unverified and should be checked out. Note, do not use a P/N with CR suffix, it will be reversed polarity.
References that were useful were the Amps Reflector Archive, which can be searched by anyone. Several members of the Amps Reflector contributed ideas and suggestions. On Yahoo there is an Alpha 87A group and those messages can be searched by a non‑subscriber. Very useful info was found on both sources.

Post written by:  Charlie, N8RR