I'm in the market for a used spectrum analyzer like the 7L12
or the HP 141-T. One important use will be to make two-tone
IMD measurements on HF SSB transmitters. I am concerned that
with tone separations on the order of one kHz, the 7L12 may
not have sufficient bandwidth in the 300 Hz mode to resolve
adjacent IMD products expected to differ in amplitude by 40
dB or more.

Has anyone on the group used a 7L12 for this purpose and is
the 300 Hz RBW sufficiently narrow?

There seems little doubt that the 141, with 10 Hz or 100 Hz
RBW, will handle this.

Many thanks in advance.

Chuck
NT3G

On Wed, 18 May 2005 19:04:44 -0400, chuck wrote:

I'm in the market for a used spectrum analyzer like the 7L12
or the HP 141-T. One important use will be to make two-tone
IMD measurements on HF SSB transmitters. I am concerned that
with tone separations on the order of one kHz, the 7L12 may
not have sufficient bandwidth in the 300 Hz mode to resolve
adjacent IMD products expected to differ in amplitude by 40
dB or more.

Has anyone on the group used a 7L12 for this purpose and is
the 300 Hz RBW sufficiently narrow?

There seems little doubt that the 141, with 10 Hz or 100 Hz
RBW, will handle this.

Many thanks in advance.

Chuck
NT3G

I have never played with a 7L12 but do have a 141t. 300hz bandwidth
will work. The best resolution on the 141t is 100hz. There is a 10 hz
video filter but the video filter does no good for resolution. It only
gets rid of noise. The biggest problem that I can imagine with the
7L12 would be whether or not your scope has some kind of storage or
long persistency on the tube. To look at 300 hz bandwidth you need a
very slow sweep speed on the scan. If there is no storage you will not
be able to see it.

73
Gary K4FMX

Thanks for responding, Gary.

Sweep time is something that I had not even been
considering, so I appreciate the info.

From what I can gather about the 7L12, the slowest sweep is
10 mSec/division, or 0.1 second for a complete sweep. That
is the sweep setting Tek says to use for all frequency
domain analyses, regardless of bandwidth. The other sweep
settings go from 10 mSec/division to something like 1
uSec/division, but they are intended for time domain analyses.

As a reality check on a 0.1 second sweep, I looked at HP's
performance test procedure for the 8552B to see what sweeps
they specify for measuring the 8552B's bandwidth. For the
300 Hz RBW, they list a 0.2 second sweep (I assume that's
not 0.2 second/division). Not really too far from Tek's 0.1
second. Of course, at the narrower bandwidths, HP's
specified sweeps get much slower, as you pointed out. And if
it is 0.2 second/division, then there is a profound
difference between the Tek and HP filter designs or I'm
missing something big. As I recall, the slower sweeps are to
avoid ringing in the filters, gaussian skirts notwithstanding.

So maybe the 7L12 doesn't require a storage scope for the
300 Hz RBW? Sure wish I had access to the Tek Op manual for
the 7L12.

73,

Chuck






Gary Schafer wrote:
On Wed, 18 May 2005 19:04:44 -0400, chuck wrote:


I'm in the market for a used spectrum analyzer like the 7L12
or the HP 141-T. One important use will be to make two-tone
IMD measurements on HF SSB transmitters. I am concerned that
with tone separations on the order of one kHz, the 7L12 may
not have sufficient bandwidth in the 300 Hz mode to resolve
adjacent IMD products expected to differ in amplitude by 40
dB or more.

Has anyone on the group used a 7L12 for this purpose and is
the 300 Hz RBW sufficiently narrow?

There seems little doubt that the 141, with 10 Hz or 100 Hz
RBW, will handle this.

Many thanks in advance.

Chuck
NT3G


I have never played with a 7L12 but do have a 141t. 300hz bandwidth
will work. The best resolution on the 141t is 100hz. There is a 10 hz
video filter but the video filter does no good for resolution. It only
gets rid of noise. The biggest problem that I can imagine with the
7L12 would be whether or not your scope has some kind of storage or
long persistency on the tube. To look at 300 hz bandwidth you need a
very slow sweep speed on the scan. If there is no storage you will not
be able to see it.

73
Gary K4FMX

chuck wrote:
Thanks for responding, Gary.

Sweep time is something that I had not even been considering, so I
appreciate the info.

From what I can gather about the 7L12, the slowest sweep is 10
mSec/division, or 0.1 second for a complete sweep. That is the sweep
setting Tek says to use for all frequency domain analyses, regardless of
bandwidth. The other sweep settings go from 10 mSec/division to
something like 1 uSec/division, but they are intended for time domain
analyses.

As a reality check on a 0.1 second sweep, I looked at HP's performance
test procedure for the 8552B to see what sweeps they specify for
measuring the 8552B's bandwidth. For the 300 Hz RBW, they list a 0.2
second sweep (I assume that's not 0.2 second/division). Not really too
far from Tek's 0.1 second. Of course, at the narrower bandwidths, HP's
specified sweeps get much slower, as you pointed out. And if it is 0.2
second/division, then there is a profound difference between the Tek and
HP filter designs or I'm missing something big. As I recall, the slower
sweeps are to avoid ringing in the filters, gaussian skirts
notwithstanding.

So maybe the 7L12 doesn't require a storage scope for the 300 Hz RBW?
Sure wish I had access to the Tek Op manual for the 7L12.

73,

Chuck

Hi Chuck,

I have used both the 141T system, and the 7L13. I ditched my 141T because
the 7L5, 7L13, and 7L18 plugins perform better than the equivalent HP plugins
for the 141T system, and are much more compact.

But,

When I was looking into 7L analyzers, I discounted the 7L12. It is too
primative. The minimum you want to do any real work is a 7L13 with a
7633 storage frame. The 7L14 is much better because it has the digital
storage, but it is also twice the price of a good 7L13.

You *will* need a storage scope frame for either the 12, or the 13. The
narrow bandwidth sweeps must be done really slowly. If you try and rush them,
you will lose most of the amplitude information... the filters just cannot
respond quickly. It is a physical reality of narrow band filters.

-Chuck Harris

From one Chuck to another,

Thanks for the info and the advice! I will definitely look
further into the '13 and '14.

I do understand the need to sweep "slowly" at narrow RBWs.
But I'm still troubled by the fact that the *slowest* sweep
built into the 7L12 is 10 msec/division! That will,
arguendo, degrade the filter response. The storage scope
will surely not sweep the SA at a slower rate, and putting a
distorted SA output signal into a storage scope can't
possibly reshape the response! So there is no cure. If our
assumptions are correct, this is a fatal Tek design flaw
(not a whole lot of them around).

A storage scope would be really important if the SA is
sweeping too slowly for the regular scope's persistence, or
to capture a single-sweep trace. Or for simply storing a
trace for later viewing. But if the sweep rate is 10
sweeps/second, there shouldn't be much flicker with P31.

Something is amiss here, I think. Maybe there is a typo in
Tek's spec sheet? Or more likely, a parity bit error in my cpu!

Chuck

chuck wrote:
From one Chuck to another,

Thanks for the info and the advice! I will definitely look further into
the '13 and '14.

I do understand the need to sweep "slowly" at narrow RBWs. But I'm still
troubled by the fact that the *slowest* sweep built into the 7L12 is 10
msec/division! That will, arguendo, degrade the filter response. The
storage scope will surely not sweep the SA at a slower rate, and putting
a distorted SA output signal into a storage scope can't possibly reshape
the response! So there is no cure. If our assumptions are correct, this
is a fatal Tek design flaw (not a whole lot of them around).

A storage scope would be really important if the SA is sweeping too
slowly for the regular scope's persistence, or to capture a single-sweep
trace. Or for simply storing a trace for later viewing. But if the sweep
rate is 10 sweeps/second, there shouldn't be much flicker with P31.

Something is amiss here, I think. Maybe there is a typo in Tek's spec
sheet? Or more likely, a parity bit error in my cpu!

Chuck

No, your data sheet is wrong. The slowest automatic sweep is 10 secs per
division. The slowest sweep is manual. This specification exists across
the entire 7L line.

The 7L5, 7L13, 7L14, and 7L18 all have monitors built in that will show
"uncal" if you sweep too fast for the filter setting.

-Chuck

On Thu, 19 May 2005 20:15:16 -0400, chuck wrote:

From one Chuck to another,

Thanks for the info and the advice! I will definitely look
further into the '13 and '14.

I do understand the need to sweep "slowly" at narrow RBWs.
But I'm still troubled by the fact that the *slowest* sweep
built into the 7L12 is 10 msec/division! That will,
arguendo, degrade the filter response. The storage scope
will surely not sweep the SA at a slower rate, and putting a
distorted SA output signal into a storage scope can't
possibly reshape the response! So there is no cure. If our
assumptions are correct, this is a fatal Tek design flaw
(not a whole lot of them around).

A storage scope would be really important if the SA is
sweeping too slowly for the regular scope's persistence, or
to capture a single-sweep trace. Or for simply storing a
trace for later viewing. But if the sweep rate is 10
sweeps/second, there shouldn't be much flicker with P31.

Something is amiss here, I think. Maybe there is a typo in
Tek's spec sheet? Or more likely, a parity bit error in my cpu!

Chuck








I just looked at the 141t and it takes 2ms/div for no flicker. This is
at wide bandwidth.

At 2khz/div spectrum width and 300 hz bandwidth it takes 50ms/div
sweep speed max. At 5 khz spectrum width it takes .1 sec/div sweep
speed.

At 100hz bandwidth and 2khz/div spectrum width it takes .5sec/div
sweep speed.
Definitely storage scope area!

The 50ms/div sweep can be viewed without storage on a p31 but it is
not good.

73
Gary K4FMX

Thanks for the additional info, Gary.

As you can see, I'm having a difficult time understanding
just what the 7L12 does. There may be a typo in the data
sheet, but when I look at the time/div dial on the 7L12, it
just doesn't have the numbers to do anything slower than 10
ms/div. The other models in the 7L series have many more
sweep speeds available, based both on their data sheets and
on photos. I suspect that the 7L12 was different. In fact,
because the 7L13 has 100 Hz and 30 Hz RBWs, it would have to
have slower sweep rates.

Your finding with the 141 was that with a 300 Hz bandwidth
and 2 kHz/div width, a sweep rate of 50 ms/div was marginal.
I imagine then that a sweep rate of 10 ms/div (assuming that
would be compatible with the 300 Hz filter) would be ok with
P31. If I understand correctly.

I think we've gone about as far as we can without either
having a 7L12 in front of us or at least an op manual.
Thanks again.

73,

Chuck

On Thu, 19 May 2005 20:49:29 -0400, Chuck Harris
wrote:

chuck wrote:
From one Chuck to another,

Thanks for the info and the advice! I will definitely look further into
the '13 and '14.

I do understand the need to sweep "slowly" at narrow RBWs. But I'm still
troubled by the fact that the *slowest* sweep built into the 7L12 is 10
msec/division! That will, arguendo, degrade the filter response. The
storage scope will surely not sweep the SA at a slower rate, and putting
a distorted SA output signal into a storage scope can't possibly reshape
the response! So there is no cure. If our assumptions are correct, this
is a fatal Tek design flaw (not a whole lot of them around).

A storage scope would be really important if the SA is sweeping too
slowly for the regular scope's persistence, or to capture a single-sweep
trace. Or for simply storing a trace for later viewing. But if the sweep
rate is 10 sweeps/second, there shouldn't be much flicker with P31.

Something is amiss here, I think. Maybe there is a typo in Tek's spec
sheet? Or more likely, a parity bit error in my cpu!

Chuck

No, your data sheet is wrong. The slowest automatic sweep is 10 secs per
division. The slowest sweep is manual. This specification exists across
the entire 7L line.


That's not what this says:

http://www.tucker.com/images/images_spec/00000453.pdf


I wouldn't assume that the resolution BW of the 7L12 would be adequate
to look at IMD down 40+ dB at 1 KHz spacing. The BW is specified as
300 Hz at -6dB with a -6 to -60 dB shape factor of 4:1. I don't have
time to plot the selectivity curve at the moment, but it might be an
exercise for you.

If you have a "perfect" cw signal as input, when you sweep it, what
you are plotting on the screen is the filter response of the SA.

Figure a Gaussian response with the specified shape factor and then
overlap two of the curves with 1 KHz spacing and see if the filter
skirts are down 50 dB where they overlap. (You should have 10 dB of
margin IMHO)

Personally I'd use the HP with a storage mainframe.

No, your data sheet is wrong. The slowest automatic sweep is 10 secs per
division. The slowest sweep is manual. This specification exists across
the entire 7L line.



That's not what this says:

http://www.tucker.com/images/images_spec/00000453.pdf

I pulled out a 1981 Tek catalog, and I am indeed remembering wrong.
This is yet another reason why I discounted the 7L12 as a credible SA.

Here is what Tek is expecting you to do. There are timebase positions
for 5ms through 0.01us. These are for use when the SA is being used as
a receiver (time domain mode), and you are looking at a received pulse
train.

The "SA" mode is *variable* from 10ms through 5ms/division. You are expected
to manually adjust the sweep rate to get a clear picture.

So, how can you get 300Hz filter resolution? Well, simply by not scanning
the full bandwidth of the SA plugin. Reduce the sweept bandwidth to 10KHz,
and 300Hz is easily achieved with a 5-10ms/division sweep rate. Not a great
way to go, but usually if you are interested in the narrow resolutions, you
are only looking for signals over a narrow bandwidth.

As I said earlier, get a 7L13, or 7L14. The 7L12 wasn't fully incubated
when it was hatched.

When you go looking at the 141T, remember, it is a mid 1960's SA design,
and it feels like it when you use it. The 7L5, 7L13, and up were designed in the
very late 1970s, and take advantage of things like microprocessors to help
with house keeping operations. They are smaller, quite reliable, and just
plain work better than the 141T family. (And, yes I have owned, used and
repaired both.) The 141T storage tube is a nightmare. Very short life.

Another SA line that is usually very inexpensive, and much better than the 141T
family, is the Eaton/Ailtech 727 and up. I was told (in the early '80s) by an
HP FAE (who specialized in HP's SA's and other RF gear) that the Ailtechs were all
over the place in HP's internal R&D labs. They were the SA's that HP used in
designing their own product line.

(flame suit on, helmet latched, as I await the onslot of rebuttles from HP guys..)

-Chuck Harris