A known problem of this rig is the noise blanker function. Some people say it do not perform well, and others say it does not work at all. I will describe my solution to this problem, tested on three different transceivers and working well on all of them. Of course this is based on my own experience and must be taken only as a starting point.
What is a noise blanker?You must to understand how a noise blanker works before modify it. The noise blanker is a device designed to reduce, or even eliminate impulsive noises, like the one produced at spark gaps, electrical motors, and so on. It detects the large amplitude short duration pulse and reduces the receiver's overall gain accordingly, in a way the pulse is not present at the receiver's output.
A noise blanker works detecting the noise pulses and pulling down the AGC in the same but reverse way, reducing the overall receiver's gain just some milliseconds (or even microseconds). In this way, the noise pulse is not present at the receiver's output.
In theory you will never need a noise blanker, but in practice, it is a fundamental device in nowadays radio, specially in urban and industrial areas.
The noise blanker implemented in the FT-817Here you can see the schematics of the noise blanker implemented in the FT-817. T1027 is IF transformer that picks up the signal before the IF filters. This is very important to get the real pulse shape. This signal is amplified by Q1052 and Q1061 and feed to Q1063, configured as a current buffer. Once amplified, this signal gets into D1068, a pair of diodes mounted as voltage doubler rectifier. This rectified signal is divided in two paths. The first trough R1343 and Q1076, making the own AGC for the noise blanker, and the other through R1334, Q1074 and Q1112, the one who acts over the main receiver AGC.
FT-817's noise blanker implementation
The signal used to feed the own noise blanker AGC is filtered with C1362, but the one used to feed the main receiver's AGC through Q1074 is not filtered at all, so it contains the noise pulses detected by D1068.
Why it does not work?If you look at the transceiver schematics, you will see many components marked with asterisks (*). This components have values not determined yet for optimum performance, so they can vary from lot to lot of transceivers. In the noise blanker, there is a resistor marked with asterisks (R1345), so I must assume the design of this circuit have not been finished.
Noise blanker adjustment procedure, one of the many erratas present in the service manual.
In the service manual, you can read the noise blanker must be adjusted injecting a signal, tuning it, and adjusting T1027 to get the minimum voltage at Q1074 base. Obviously, this is not correct. It is one of the many erratas in the service manual. The adjust must be made to get the maximum voltage on Q1074 base.
This is the first problem you will get. In the first FT-817 version (not the ND version) you will get the maximum voltage with the T1027 core fully inside the coil, so you will never know if you get the right point or not. To see the real maximum peak, you would need more inductance!
Noise blanker components
KA7OEI has a real good page about the FT-817 ( http://www.ussc.com/~turner/ft817pg.shtml ) In his page, you can read how KA7OEI solved this problem: He soldered another small chip capacitor (about 5 or 10 pF) in parallel with R1246. In this way, you will be able to peak T1027 without problems.
The first FT-817 I modified worked nicely tuning only T1027 at its maximum value. In this way the noise blanker worked just fine.
The second FT-817 I modified did not work properly adjusting only T1027. I soldered a small 10 pF SMD capacitor just over R1246. Then, T1027 peaked without any problem.
The third FT-817, this time a FT-817ND, was more problematic. T1027 was perfectly tuned, but the noise blanker did not work at all. I measured a maximum voltage on Q1074 base of about 160 mV. This voltage is very small to polarize Q1074, so the noise blanker can't work at all. But at D1068 output, voltage was about 550 - 600 mV, so now the problem was perfectly located: The own noise blanker's automatic gain control (AGC) regulated its output to a very low value (this is the point where R1345 gets in), so Q1074 can't work properly.
R1345 is located at the solder side, not very accessible indeed, so I played with the voltage divider formed by R1334-R1333, much more accessible. I removed R1333 and all I got was a excessive actuating noise blanker: It distorted heavily CW and AM signals.
This was good, because it means there was enough voltage at D1068 to make the noise blanker work. I soldered a 47K multiturn variable resistor at R1333 place and I spent a whole evening looking for the optimal point for the noise blanker.
A 47K multiturn resistor, and two pieces of wire can help you to find the proper value for R1333
Basically, if there is enough voltage at D1068 output (with input signal!) to polarize Q1074, R1333 regulates the noise blanker actuation. I found a 25K resistor worked fine for me, so I replaced R1333 with a 22K SMD resistor and all worked fine.
Repair guide:1.- Check T1027 tuning. Sometimes adjusting this coil will makes the NB to work properly.
2.- Check voltage at the D1068 output (accessible at R1345). If voltage in presence of signal is not at least 500 mV, you will need to modify R1345 to get such or greater voltage.
3.- Adjust R1333 to get the noise blanker to work properly.