Wednesday, March 1, 2017

Looking for ALC level problems in HW-100


One issue I have had is lack of any motion on the meter in Relative Power mode, which turns out to be driven by circuitry right next to the components that develop the negative ALC bias voltages that will lower the amplification of the isolation amplifier V2 and the first IF amplifier V3. 

My guess on the relative power meter is that I am not delivering a strong signal to drive the final tubes, thus there actually is virtually zero power. This is suggested by two data points - lack of any heat from the dummy load hooked to the antenna, and lack of change to the plate current.

This developed ALC voltage is used when in LSB or USB mode, but the MIC/CW gain control sets the bias voltage for CW ot Tune modes. I can see the voltage from the control swing from -10 down to 0 as the control is advanced. I must look at the ALC lines and voltages more thoroughly.

To recap, I know that the carrier oscillator is producing its beat frequency oscillator (BFO) output and that this is delivered to the balanced modulator circuit. I know that audio is delivered to unbalanced the modulator and that it is switched to unbalanced mode while in CW or Tune.

The resulting signal is reaching the transformer that delivers this modulated BFO to the isolation amplifier V2. I don't see the signal amplified by V2, which blocks it from reaching the IF amplifier and moving on to the first transmitter mixer where it will be mixed with the tunable VFO that selects specific frequencies for reception or transmission. 

V2 and the first IF amplifier V3 are tied to the ALC voltage, such that the amplification is reduced when high power is detected in the final amplifier tubes V8 and V9. It appears that the circuitry is forcing the tubes to nearly zero gain, incorrectly, since we are not generating any real power to the antenna. 

The gain is forced down by two separate mechanisms - the ALC voltage from the finals when in USB or LSB mode, and the MIC/CW gain control when in CW or Tune. I can't produce any relative power regardless of the mode. 

V2 is a grounded grid amplifier, meaning the grid closest to the plate (suppressor) is at ground, the BFO frequency is injected as AC swings on the cathode, and the screen grid is at 120V while the plate is at 150. Thus, the electron flow from cathode to screen grid is controlled by the bias voltage on the control grid, which ranges between 0V at minimum amplitude to -10V when the gain is cranked up. 

The bias voltage is derived from a junction with two resistors, a 47K and a 1M value. When I measured the voltage across the 47K, it was spot on, but when measuring across the 1M resistor I have a quite low value, nowhere near 1M. If there is a legitimate resistance in parallel with the 1M resistor, it renders it effectively redundant. This leads me to conclude that the resistance should NOT be this low and something downstream from this line is bad.

Resistor junction with one seemingly bad measurement
I can't imagine how a 1M resistor would drift downwards that much in value, thus I have to look at all the connected components to find which part(s) are off value. I looked at the "GRN" line which comes off the bottom of the 1M resistor and traced it back to capacitors that should isolate it entirely at DC.

One section of the GRN line with isolating capacitors or grids
This originates in the final amplifier section, but is DC isolated, so I moved back towards the 1M resistor and found a short segment that does conduct at 10K but only if there is a leaky or shorted capacitor or tube in the above section. Without the leak.short, the 10K resistor cannot affect the value of the measurement across the 1M component.

Coupling but only if the section above has leaks or a short

Going further back, the GRN line does go through some resistors that are eventually headed to ground, but the other side of the 1M resistor would also need to get to ground in order to have a low parallelled resistance. I measure about 58K ohms so the path I am seeking is not going to be much above that since the 1M contributes very little to dropping the composite resistance.

One path for the GRN line is through the mode switch to supply a bias to the Heath built VFO when in LSB mode, to shift the frequency to the other side of the BFO (carrier) frequency. It is disconnected because I replaced the Heath VFO with the direct digital synthesizer.  We can ignore this.
Irrelevant path to VFO bias using wht-gray wire, which is disconnected
The next section of the GRN path is the source of the negative  bias voltage, which comes from the HP-23 external power supply and our value of the voltage on GRN is the divider with BIAS potentiometer. This sets our negative bias somewhere between about -40 and -130V on the GRN line. 

Divider to deliver negative 
For the 1M resistor to be paralleled down to 58K, I would need something, on the path leaving the 47K resistor at the junction with our 1M part, that has a resistive path to the negative bias line with roughly 12K of resistance. That is the only way we can force the 1M value down to what we measure.

The path from the 47K resistor goes to V7 the driver tube, where it should be isolated unless the capacitor or grid is leaky/shorted.
Isolated section from 47K resistor unless leaky/shorted
The line also goes to the grids of the final amplifiers, but in a way that should be isolated at DC unless we have a leaky or shorted capacitor or grid. Nothing in the schematic shows a route that should connect the path from the original 47K resistor with about 12K of added resistance into the GRN path.

Isolation unless caps or tubes are leaky/shorted
I am going to have to interrupt various parts of the circuit and find the low resistance path that must exist, or my 1M resistor itself has plummeted in value. This will require some disconnections and resoldering - going to be a slow process but I need to find and correct this issue in order for the transmitter section and receiver ALC to work properly.

The 1M resistor itself checked out fine, so the reality is what I expected, that paralleled resistance is pulling down the effective resistance. More investigating ahead.

No comments:

Post a Comment