Hammarlund HQ170 Restoration
|
 |
|
Although I refer to this
exercise as "restoration" I intend to do the minimum
amount of work to get this old receiver running tolerably well.
I remember trying it out many years ago when I first acquired
it and was surprised how sensitive it was and how quiet was background
noise.
The first step was to lug out
my 240/110 volt isolation transformer for which I paid £1
at a street auction donkeys years ago but when I plugged it in
the was a dull thud and a cloud of smoke smelling like burnt
rubber. After ditching the input and output cables and refitting
with a new plastic covered output cable and adding an IEC connector
on the front it worked a treat and the HQ170 dial lamps lit up.
Next, let's see exactly how
the receiver performs..... most disappointing because after connecting
an 80m dipole or a long wire all I was able to hear across its
seven bands was a couple of very weak CW stations and a few weak
teletype noises. I switched on my newly repaired Wavetek 2407
(with some trepidation... but it worked fine) and checked the
receiver calibration and sensitivity (or as I was to discover:
It's degree of extreme deafness).
During testing I found it very
awkward to read the dials because the number of KHz between adjacent
dial markings varies from band to band. I suppose that's something
a user would get used to.
The next day I spent a
little more time on this receiver. As it uses an inconvenient
tagstrip for the aerial I removed a wander plug socket fitted
in the past by a previous owner, drilled out the hole slightly
and fitted a BNC socket. That will provide a more reliable connection
to my signal generator. |
 |
|
I also noticed the 100KHz calibrator
crystal wasn't gripped in its socket so I squeezed the socket
and this was better. Once done I plugged in the set and connected
my 80m dipole. Much to my surprise the set was now pretty sensitive
and lots of signals were present on the 80m metres and above.
All I can think is some valve pins may be making better contact
than during the first test I did, perhaps disturbed by drilling
out the hole for the aerial socket? SSB signals on 20m were hitting
S9 +20dB on the S-meter, so not so much work to do as I first
thought.
Before repeating a set of calibration
and sensitivity checks now that the receiver's sensitivity is
OK (initial tests on all bands needed some tens of millivolts
to produce an S9 reading.. see later), I sorted out a discrepancy
between the two dials. When the left dial was set against its
cursor in its centre position I saw the right dial scales were
out by about an inch. I eventually found an Allen key that fitted
the grub screws and managed to twist the dial to match the one
on the left (the two dials are linked together by pulleys and
cord with just the one tuning condenser).
Below, the results... only a
little tweaking required on most of the bands, although 6m looks
ropey. |
|
Dial reading MHz |
Signal in MHz |
Signal in for S9 |
|
Dial reading MHz |
Signal in MHz |
Signal in for S9 |
|
1.803 |
1.800 |
64uV |
|
2.000 |
1.993 |
64uV |
|
3.513 |
3.500 |
15.7uV |
|
3.811 |
3.800 |
15.8uV |
|
7.008 |
7.000 |
15uV |
|
7.250 |
7.240 |
16uV |
|
14.000 |
14.002 |
30uV |
|
14.300 |
14.304 |
30uV |
|
21.000 |
21.144 |
40uV |
|
21.400 |
21.558 |
40uV |
|
28.000 |
28.021 |
22uV |
|
30.000 |
30.056 |
45uV |
|
50.000 |
49.977 |
428uV |
|
54.000 |
53.976 |
328uV |
|
|
Before starting the tests I
adjusted the S meter sensitivity to nearly maximum. Once I've
completed alignment I'll set it to read S9 at 50uV which is the
generally accepted signal strength for S9 on the HF bands. I
turned the RF gain to full, selected AM and switched to "Both
Sidebands" and Slot Frequency fully anti-clockwise. I left
the Antenna tuning where it was but of course adjusting this
for each reading may have improved the receiver's response.
While the signal generator was
connected I also looked at the approximate settings of the IF
amplifiers. With the 50MHz band selected a 100mV signal at the
first IF of 3.035MHz produced the following...
The S-meter reading rose as
the test signal increased in frequency with S3 being achieved
at 3.03580MHz. As the frequency was increased further the S-meter
reading rose then fell back so that S3 was achieved at 3.03587MHz.
The average frequency being about 3.03584MHz. It should be 3.035MHz.
Setting the HQ170 to top band
I checked the 2nd IF. I reduced to RF input to give me an S3
reading from 453.0KHz to 458.7KHz which corresponds to an average
frequency of 455.85KHz. A peak reading was achieved at 455.2KHz.
This gives a spread of -2.2KHz to +3.5KHz. Not too bad but could
be better.
Twiddling the various controls
told me a few things that will need correcting. First, the audio
control doesn't increase the volume when turned beyond the third
mark (I'll check for a positive bias on the 6AQ5 control grid
or just replace the audio coupling condenser. I may also test
the 6AQ5). Note: I later found that the design of the volume
control circuitry is a bit odd in that, once reaching less than
half-way, it alters some of the parameters to improve audibility
without making the audio any louder. I also found the background
noise when switching between SSB Lower and Upper sounds a lot
different so the BFO frequency may not be on the correct frequency.
The design of the HQ170 was
quite advanced for its time and uses a product detector for SSB
reception. This works well on 20m even with the RF gain at full
but it needs the RF gain reducing for less strong signals on
80m to avoid distortion. The noise limiter inner pot isn't connected
(in fact the circuit diagram doesn't show a pot here??) and,
although the calibrator worked fine when I first turned on the
receiver, the crystal socket still isn't very reliable.
Changing the AVC switch setting
alters the overall noise level so there may be something leaky
on the AVC line? In fact there may be quite a lot of work to
do to bring the receiver up to my liking.
I'll adjust the frequencies
at the top and bottom of each band to match the dial calibration
then repeat the sensitivity tests.. or should I initially check
the IF tuning? Maybe confirm the IF tuning is correct first in
case one or more of the settings is out otherwise I may need
to repeat the local oscillator tuning? My best bet is to use
my spectrum analyser to view the IF response shapes and centre
frequencies, but then again maybe not as the manual seems to
provide a recommended method based solely on twiddling against
a meter. |
|
|
|
Not yet realising the
reason for the odd behaviour of the volume control, I tested the 6AQ5 output valve and found it
was marked "Foreign" and gave me a reading of 80% on
my AVO valve tester, so I checked its grid and found it sitting
at only 6mV which is fine. The grid coupling component (above
left) turns out to be a hybrid module carrying three capacitors
and two resistors. Checking the 6AQ5 cathode I discovered its
decoupling capacitor was a third element of the main aluminium
can carrying the reservoir and smoothing condensers. Disconnecting
the 40uF part gave me a reading of 60uF and an ESR of 1 ohm which
is not ideal but still OK. HT read 263 volts and 243 volts at
the reservoir and smoothing condensers which is quite acceptable.
The second hybrid (above right)
is used in the crystal calibrator circuit.
An observation before going
further. The HQ170 uses a final IF of 60KHz. Because of this
low frequency and because several stages of amplification are
used, it is essential that circuits are not tuned by just peaking
everything otherwise you'll end up with miles too much gain and
a bandwidth really too narrow for comfortable listening to AM.
I remembered this after carrying out the first IF alignment,
then re-read the manual where it defines the receiver settings
for 60KHz alignment. I then recalled tuning up an early superhet
from 1935 and finding the response to be so narrow as to be virtually
unusable. Sometimes a user manual will tell you how to stagger-tune
the IF transformers, but with the HQ170 the proper settings and
peaking seems to work reasonably well. I did observe some double-humping
when tuning across strong signals so I may return to the job
but using my spectrum analyser.
I decided to investigate the
small discrepancy in the 3.035MHz IF and found that adjustments
are made via the bottom slugs in a pair of cans. Fortunately
I found a suitable tuning adjuster in my collection of such things
(see notes near the end of this page) and made a small correction.
Then I found that there was an intermittent, apparently inside
one of the two cans, which resulted in the extreme deafness I'd
noted when first turning on the receiver. I jiggled the core
in the top of the can and restored normal working, but that fault
will need fixing.
I spent an hour listening to
the various amateur bands and found the receiver to work very
well. It has a double-sideband setting for SSB, something I've
not seen before, which produces very good results. A harsh LSB
signal can sound much mellower in the double setting. Oddly the
LSB/USB/Double switch affects AM reception. Maybe this a design
feature?
After tuning around and listening
for a while I noticed the right hand dial had slipped back so
the marked band edges were now again an inch out. I wonder if
one of the grub screws is loose but to check these means making
a special tool because my only Allen key is too short and when
extended goes in at an angle and is awkward to turn. The dials
are coupled by a steel wire wrapped around pulleys. For the moment
I don't think it advisable to carry out alignment until the dial
slippage is fixed. |
|
|
|
Below, the now defunct clock.
This would have been designed to run on US 60Hz mains so would
run slow on UK mains. Probably the best way to restore this is
to ignore the switching arrangements at the rear, remove everything
behind the face and fit a modern electronic unit, although finding
a 24 hour type at a sensible price might be difficult. As I don't
wish to leave the power on all the time the original purpose
of the clock to turn on the receiver for 30 minutes before use
is superfluous anyway. |
|
|
|
I decided to carry out
more tests as I'd noticed a few problems that would interfere
with alignment. There are a couple of intermittent problems,
one of which is a rise or fall in background noise level. This
is triggered by touching any part of the circuit near the 2580KHz
crystal which is soldered in place below the oscillator valveholder.
At first I thought it was something inside one of the cans then
the crystal or a dry joint around the valveholder, but it seems
to be a loose or shorting connection inside the transformer can.
I made a sketch of the pin wiring
at the suspect T2 transformer can, unsoldered everything and
unclipped the can from the chassis. A U-shaped spring with tiny
teeth holds the can in place. The coil former looked a bit scruffy
and I soon discovered it had been removed once before. There
are three coils wired to six pins which are located in a ceramic
base. At some time part of the ceramic material has been broken
off and araldite now holds two of the pins in place. First, I
cleaned off old solder then cleaned up the wire ends and resoldered
all six wires. The thin wires to the coils looked bare in places
so I fitted a paper sleeve inside the can to prevent a short.
I also noticed two of the pins could be bent to touch the can
surround so I fitted tape around the bottom of the coil to cover
where the ceramic insulation was missing. Refitting the transformer
and tapping it gently indicates the intermittent fault is now
cured.
A really annoying thing is the
manual for the set refers to transformers by their circuit reference
number, but there isn't a drawing showing these numbers (later
HQ170A documentation is much better in this respect). The previous
owner has marked the 60KHz cans but all the rest are anonymous,
so I produced the drawing below.
Yesterday I'd (inadvertently)
peaked all the 60KHz transformers, thinking this was what was
intended, but found the receiver gain was too high, to the point
of instability, and the bandwidth switch settings were most peculiar.
I recall the slugs needed a lot of movement and thought this
rather strange at the time so I then re-read the manual and found
the stages are stagger-tuned by extra components wired via the
bandwidth switch. Correct adjustment needs to be made by setting
the receiver to AM, switching to lower sideband and 0.5KHz bandwidth
and then peaking them all. Sure enough, once done I found the
gain was manageable and the switched bandwidth control results
sounded much better.
Nearly all the slugs in the
various transformers require a hex shaped plastic trimming tool
and most of these require the hex shape to be at the end of a
necked section allowing the bottom slugs to be set from the top
of the chassis. For some reason the exception to this are the
three 455KHz transformers at the front of the chassis. For these
I had to file a trimming tool from an old plastic knitting needle
to match the slot in their slugs (see the note near the bottom
of this page). I can now adjust the top slugs but those at the
bottom won't adjust (I suspected the cores had been damaged).
Because of this the tuning of the 455KHz IF strip is about 1.5KHz
too high.I suppose, if all else fails, I could add a little capacitance
to bring down the setting?
I've also noticed that the RF
from the final 60KHz IF strip is crackly. Looking at the output
on a scope it's jumping around. Maybe a capacitor is acting up
or there's a dry solder joint? I didn't get to the bottom of
this problem and it's possible the set is so sensitive that it's
picking up local electrical interference which is getting into
the 60KHz IF strip and modulating incoming signals. |
|
Below, a drawing of the
valve and coil locations marked up with their circuit part numbers.
Top=front of set.
Below this drawing is a chart
giving coil locations within the various cans. |
 |
|
COIL |
ADJUST-KHz |
ADJUST-KHz |
|
COIL |
ADJUST-BAND |
ADJUST-BAND |
|
NUMBER |
TOP |
BOTTOM |
|
NUMBER |
TOP |
BOTTOM |
|
L4 |
395 |
- |
|
T12 |
160 |
- |
|
T1 |
455 |
3035 |
|
T13 |
80 |
- |
|
T2 |
455 |
3035 |
|
T14 |
40 |
- |
|
T3 |
455 |
455 |
|
T15 |
20 |
- |
|
T4 |
455 |
455 |
|
T16 |
15 |
- |
|
T5 |
455 |
455 |
|
T17 |
10 |
- |
|
T6 |
60 |
60 |
|
T19 |
160 |
80 |
|
T7 |
60 |
60 |
|
T20 |
40 |
20 |
|
T8 |
60 |
60 |
|
T21 |
15 |
10 |
|
T9 |
60 |
60 |
|
T23 |
160 |
80 |
|
T10 |
60 |
60 |
|
T24 |
40 |
20 |
|
T11 |
60 |
60 |
|
T25 |
15 |
10 |
|
T28 |
60 |
- |
|
- |
- |
- |
|
|
I can't stress too much
to anyone contemplating an alignment that the correct tool types
or exact equivalents are used. Hammarlund quote two types. One
is a type 5097 which is a stubby flat blade used for T3, T4 and
T5 and the other a type 8282 which is a hex shape on the end
of a stalk. This enables the tool to adjust bottom slugs from
the top of the set without interfering with slugs on the top.
This isn't always easy and a bit of cautious jiggling may be
necessary.
The hex tool type 8282 has a
thickness between opposite flats of 2.54mm which I reckon is
supposed to be a tenth of an inch. The length of the hex section
needs to be around 10mm and the necked part about 2mm in diameter.
A length of 100mm is OK. I had one of these in a cheap kit of
alignment tools.
I filed a knitting needle to
the exact fit for tool 5097 and it ended up 3.8mm in diameter
with a blade width also 3.8mm and a thickness of 1.9mm. The length
of the flat part need only be no more than 3mm to fit the slot
in the slugs. The underside slugs need the tool to be no more
than 3.8mm in diameter because the entry hole diameter is smaller
than the top hole. If you use a 4mm plastic rod this needs to
be reduced over about 25mm to reach the deepest slugs. A length
of 100mm is OK. |
 |
|
I found it a bit tricky
to adjust the slugs because some were quite sticky in the threads.
I suppose there's a chemical or oil that could be used to free
them but I hesitated using anything in case it resulted in something
nasty happening such as altering the basic coil inductance.
I spent a little more time on
alignment once I'd made a decent 455KHz trimming tool and found
one slug quite a way out. What had caught me out was the holes
in the bottom of the cans are smaller than those at the tops
and jamming the adjusting tool. Once I'd reduced the diameter
of the tool the slugs tuned up easily. Finally, I was happy with
results. During the job I'd noticed the controls for the BFO
tuning and Slot Frequency which are both small variable condensers
were slightly marred by protruding heads of the securing screws
(are these original?) so I removed them together with their star
washers and fitted countersunk types. Note that the two fixing
screws for the Slot Frequency control need to be quite short
otherwise they'll protrude too far and foul the condenser vanes.
Another tip concerns adjustment
of the Slot Frequency coil. Mine was very stiff and was impossible
to tune precisely. I discovered that if the pointer didn't quite
line up to the zero mark the pointer can be pushed slightly to
line up accurately.
The tuning mechanism is of fairly
poor quality although it works well enough in practice. If you
notice some backlash, this can be minimised by tightening the
nut at the rear of the flywheel but at the expense of spin-ability.
I don't know if a brand new receiver flywheel worked well enough
to spin the dial, but I found that this isn't possible with a
well used example. In practice a slightly heavy tuning knob is
OK. Fine tuning can be carried out using the fine tuner which
is ideal for tracking an SSB signal when either the station or
the receiver frequency is drifting.
Below, waiting for a clean-up
of the front panel, then I'll carry out a sensitivity check to
see if there's been any significant improvements, then if all
is well, fit the set back in its case and put it back on display.
Before I did this, one of the controls had been annoying me.
The Send/Receive/Cal switch had lost its positive ident. I dug
out my box of ball bearings thinking one had fallen out of the
yaxley switch but found it used only a bent metal wiper to engage
in mouldings in the metal mounting plate. The wiper was no longer
pressing down on the mouldings but I found an easy way to fix
this. I tapped gently on the end of the metal wiper using a small
hammer against a flat-bladed screwdriver and the wiper bent inwards,
perfectly pressing against the mouldings. Now the control clicks
into position. |
 |
|
Here's a recap on receiver
performance before starting realignment |
|
Dial reading MHz |
Signal in MHz |
Signal in for S9 |
|
Dial reading MHz |
Signal in MHz |
Signal in for S9 |
|
1.803 |
1.800 |
64uV |
|
2.000 |
1.993 |
64uV |
|
3.513 |
3.500 |
15.7uV |
|
3.811 |
3.800 |
15.8uV |
|
7.008 |
7.000 |
15uV |
|
7.250 |
7.240 |
16uV |
|
14.000 |
14.002 |
30uV |
|
14.300 |
14.304 |
30uV |
|
21.000 |
21.144 |
40uV |
|
21.400 |
21.558 |
40uV |
|
28.000 |
28.021 |
22uV |
|
30.000 |
30.056 |
45uV |
|
50.000 |
49.977 |
428uV |
|
54.000 |
53.976 |
328uV |
|
|
And here's the results
after realignment |
|
Dial reading MHz |
Signal in MHz |
Signal in for S9 |
|
Dial reading MHz |
Signal in MHz |
Signal in for S9 |
|
1.800 |
1.800 |
400uV |
|
2.000 |
2.000 |
400uV |
|
3.500 |
3.500 |
50uV |
|
3.800 |
3.800 |
67uV |
|
7.000 |
7.000 |
7uV |
|
7.300 |
7.300 |
7uV |
|
14.000 |
14.000 |
13uV |
|
14.400 |
14.400 |
10uV |
|
21.000 |
21.000 |
10uV |
|
21.600 |
21.600 |
10uV |
|
28.000 |
28.000 |
8uV |
|
30.000 |
30.000 |
11uV |
|
50.000 |
50.000 |
5uV |
|
54.000 |
54.000 |
10uV |
|
|
The oscillator tracking
is now perfect so I could make measurements at the dial ends.
A few puzzling things.. why should top band sensitivity and 80m
have dropped off? These two bands are in the double-superhet
range whilst the others are all triple-superhet ranges, so it's
possible one of the 455KHz transformers is now off frequency?
Before I did this I decided
to remove the old mains lead which was a twin wire cable added
by the last owner and instead fit an IEC chassis mounted socket.
I then modified my 110 volt transformer by adding in place of
its output cable, a US-style chassis mounted socket. This enables
me now to use one of the dozens of IEC/US plug leads that came
over the years with computer equipment. I also fitted to the
rear of the receiver chassis, a phono socket wired in parallel
with the loudspeaker screw terminals. I remembered to write on
the chassis at the side of the new socket "110 V".
When I have time I'll use my
Rigol spectrum analyser to see the shape of the receiver's response.
I finally got round to doing this and found the 60KHz IF strip
was in reasonably good order except it was set to between 61
and 62KHz, so proceeded to make adjustments. After shifting the
response to centre on precisely 60KHz I made the records shown
below. The RF probe was connected to the detector diode input
whilst RF input was made to Pin 7 of the final mixer V5. Note
that the instruction manual suggests that the rectified output
is monitored. This will be a DC voltage and fine for aligning
the IF amplifier with a steady RF input but of course no use
for checking with a spectrum analyser/tracking generator. Because
my probe introduces a very large attenuation of the signal, it's
only the shapes that are important. |
 |
|
Above is the 60KHz IF
response using the narrowest bandwidth setting of 0.5KHz on the
HQ170, spread over 58 to 62KHz and below the same trace but over
a sweep of 50-70KHz. The response is exactly centred on 60KHz.
Before correcting the centre frequency with the spectrum analyser
I found it was centred on 61KHz. The horizontal markings for
top curve are spaced 400KHz apart and those for the bottom curve
are 2KHz apart, giving a response of about 700Hz at the 3dB points. |
 |
|
Below, the response of
the 60KHz IF amplifier with the bandwidth switch set to 1KHz
(actually this is specified as 1KHz SSB and 2KHz AM) and with
the sweep at 50 to 70KHz. The horizontal grid markings are now
2KHz apart. Note that the shape of this curve, and those for
the following broadening settings, is defined by dozens of small
condensers introduced by the bandwidth switch. I'm sure the curves
could be cleaned up a little by fiddling with the values of these
additional parts, but that would not be a "restoration"
but more a "redesign" project and not the object of
this exercise. The curve shows a response of about 2KHz at the
3dB points. |
 |
|
Below, the response of
the 60KHz IF amplifier with the bandwidth switch set to 2KHz
(specified as 2KHz SSB and 4KHz AM) and with the sweep at 50
to 70KHz. The curve shows a response of 4KHz at the 3dB points. |
 |
|
And finally, the response of
the 60KHz IF amplifier with the bandwidth switch set to 3KHz
(specified as 3KHz SSB and 6KHz AM) and with the sweep at 50
to 70KHz. The curve shows a response of about 6KHz at the
3dB points. |
 |
|
Note that all the 60KHz
scans were all done with the mode switch set to AM and double
sidebands selected. The shape of the final curve explains the
double-humping response as the receiver is tuned across a strong
AM station. As I mentioned above... the actual shape of the curve
is defined by small condensers wired in place by the bandwidth
switch and any attempt to modify this by retuning the transformers
would throw out the responses at other switch settings. |
 |
I found this chart showing the
60KHz response curves for an HQ170A which was a slightly improved
version of the HQ170.
The scales are very different to mine
because I used a probe which heavily attenuated the signal.
The narrow bandwidth is specified for
upper sideband whilst I used double sideband to get the pictures
above. My curves show about 1.3KHz at 10dB down which works out
at half this (about 650Hz) for single sideband.
If you look at the selectivity curves
taken from this chart you'll notice that the figures for switch
settings A, B, C and D equate to 0.5, 1, 2 and 3KHz at the -10dB
points. |
|
After the low frequency IF strip
had been adjusted the 395KHz local oscillator was set to this
exact frequency, the 455KHz amplifier peaked, then the second
fixed frequency oscillator was set to exactly 3035KHz and the
seven wave bands aligned. Next, I'll attempt to measure the performance
to see how this has changed since alignment. Connecting a signal
generator to the aerial input and an audio power meter to the
loudspeaker output terminals revealed I could tune across a signal
of 0.4uV on the 40m band to produce a well defined peak of something
like 200uW of audio.
Rechecking the performance showed
that the sensitivity for most bands had slightly improved. Those
relying on only double conversion rather than triple conversion
had worsened slightly because 455KHz adjustment was awkward,
but I found a really annoying problem and this was some bands
were no longer tuning correctly. This is down to the adjustment
of the oscillator cores. Working naturally lower to higher, adjustment
of the bottom cores had affected the top cores. Firstly due to
interaction of the two cores per coil former and secondly, as
the trimming tool is extracted the upper core is moved slightly.
Core adjustment is extremely sensitive and, because of friction
between the core and its former, the frequency changes in jumps
rather than changing smoothly. This is not easy to deal with
because of the shape of the adjusting tool makes it too springy.
The 6m band sensitivity is only a rough figure because as the
antenna trimmer is adjusted you get instability, a recognised
problem with this receiver.
|
Band |
160m |
80m |
40m |
20m |
15m |
10m |
6m |
|
Signal for S9 |
500uV |
55uV |
7uV |
15uV |
7uV |
7uV |
15uV |
Air testing showed the HQ170
to be remarkably sensitive and the various features, including
its noise limiter and notch facility were very good. Bandwidth
settings work well and a useful feature is double sideband which
can improve signal fidelity. Fine tuning of SSB signals is easy
if the fine tuner is used. This adjusts the setting of the 2nd
local oscillator rather than the main local oscillator and enables
it to be accurately calibrated in KHz from the main tuning dial.
Before leaving this note on
HQ170 alignment it's worth pointing out that the method used
is not the method used universally for the alignment of a communications
receiver. Many receivers employ a crystal filter to define their
basic response and if this is within, let's say a 465KHz IF amplifier,
alignment would be based not on a precise centreing at 465KHz,
but centred on the crystal filter response. This may be for example
466KHz and therefore this should be the frequency on which alignment
is based otherwise a significant reduction in overall gain will
result. A good example is the R206
receiver. |
|
|
|
|
