The 400Hz filter is a HP filter. If it were first followed by the LP filter the LP filter would cut off the noise from the 400Hz HP. The order HP put the filter in don't offer this quality.
The post amplifier has a gain of 100 that amplifies the noise from the filters
If you move the filter to the output of the post amplifier the total system noise will drop
considerably by the post amplifier's gain. The system SNR will improve because the signal will be well above the noise of the filters.
Shibasoku put most of the gain in the 725 before the filters.
What if you just turned the filter upside down?
Never mind, It would still need to be rewired.
Even if you start moving J200 pins around it is still
wired on the board.
Let me think (sometimes hard for me to do.)
Richard, we gotta watch out for you. You are like
a pit bull, once you get onto something I see that
you just go go go and don't stop.
Never mind, It would still need to be rewired.
Even if you start moving J200 pins around it is still
wired on the board.
Let me think (sometimes hard for me to do.)
Richard, we gotta watch out for you. You are like
a pit bull, once you get onto something I see that
you just go go go and don't stop.
Last edited:
Richard,
Where is the best place to pick from?
R14, C15, R12 junction?
to P200/p7.
Then, P200/p0 to
Pull up point, C16
from Junction of R14, C16.
Three wires total.
1. Filter bypass to TP5
2. Filter input, U4 x-point to P200/p7
3. P200/p0 to C16 up.
Pull those pins from the J200 connector to P200 board.
If I understand it correctly.
David, good explanation.
Cheers,
Where is the best place to pick from?
R14, C15, R12 junction?
to P200/p7.
Then, P200/p0 to
Pull up point, C16
from Junction of R14, C16.
Three wires total.
1. Filter bypass to TP5
2. Filter input, U4 x-point to P200/p7
3. P200/p0 to C16 up.
Pull those pins from the J200 connector to P200 board.
If I understand it correctly.
David, good explanation.
Cheers,
Last edited:
I bread-boarded the TL074. Using one amp.. check the noise on output with + input terminated in 30K Ohm. It is plenty quiet. Because the filters are working on a 30mV signal level, you throw away 30dB of range right there (compared to 1 v).
THx-RNMarsh
THx-RNMarsh
Richard and David are absolutely correct -- the filters badly need to follow the gain stage. Hard to believe HP messed that up. I'll go back and gander at the circuit to see if there is something I've missed for why they did it the way they did...
I must have been good this year as 2 VCR2Ns from HP
showed up in my mail box today.
An early Christmas.
They must have taken the long route to get
here as there were a few "delivered to wrong address"
notes hand written on it.
Can't wait to get them in and see how it does.
Cheers,
showed up in my mail box today.
An early Christmas.
They must have taken the long route to get
here as there were a few "delivered to wrong address"
notes hand written on it.
Can't wait to get them in and see how it does.
Cheers,
Got the new JFET installed and tested.
Same as the other JFET that I had installed.
Now I can put that one back in the original board.
I got 6dB more distortion on the bench than
I do in the room with the computer.
I went through the same set level procedure with this
JFET as I did with the prior JFET.
Maybe this is a good point to try some different buffer and see what they do.
Same as the other JFET that I had installed.
Now I can put that one back in the original board.
I got 6dB more distortion on the bench than
I do in the room with the computer.
I went through the same set level procedure with this
JFET as I did with the prior JFET.
Maybe this is a good point to try some different buffer and see what they do.
Its not quite so simple. if you are measuring a noisy communications channel (where the money comes from for these products) the out of band noise can overload the system and even prevent the notch from nulling. On the 725D (maybe the 725C) and on the AP2700 the 20 KHz low pass filter is pre notch for exactly this reason. The 60 Hz or 48/96 KHz stuff can be pretty high, in some cases close the the max of the system dynamic range. In those cases removing the noise is essential prior to the notch filter.
Start with your Viktor oscillator. its essential that you have a really clean source to start with. Get the analyzer optimized with it. You should be able to get below -100 dB by simply adjusting/calibrating the box. Then you can address the oscillator half of it.
I am still suspicious of the AGC stuff getting to the gate of the AGC fet. If you can see the stuff with a scope its probably way too high. I would think it should be well under 1 mV. One trick for finding the source of the distotion in the oscillator is as follows:
Use the QA400 to identify the harmonic distribution of the distortion. Then look at the output of each opamp driving the gate (there are 2 I think) and see which has the same distortion spectrum. Use the "divider" probe) we cooked up to stay inside the working range of the QA400. Clip leads for probes should be good enough.
RE noise from the filter amps -- I went back and looked at the 339's signal flow. Demian makes a very good point about using LP filtering to keep out-of-band- stuff out of the notch filter, etc, but that's not what HP did in this unit. The filters are post the input and notch stages and have a nominal 10mVRMS (at full scale) input if unit is properly cal'd.
So moving the filter stages to behind the X100 A2U4 amp and ahead of the A2U5 RMS chip is electrically simple (but may be a little fussy mechanically). This would put the input level to the filters at 1VRMS, a much better place for them to operate unless the signal is outrageously noisy -- wouldn't be good, I guess, on a telecom production line, but for us audio and instrumentation geeks, no problem -- if we keep our eyes open and our wits about us.
I plan to do this to my 339, right after I track down the 2nd H spike in the input and notch amp circuits. Let's see, I'm 74 now, so this should be finished by... um... er... heh.
So moving the filter stages to behind the X100 A2U4 amp and ahead of the A2U5 RMS chip is electrically simple (but may be a little fussy mechanically). This would put the input level to the filters at 1VRMS, a much better place for them to operate unless the signal is outrageously noisy -- wouldn't be good, I guess, on a telecom production line, but for us audio and instrumentation geeks, no problem -- if we keep our eyes open and our wits about us.
I plan to do this to my 339, right after I track down the 2nd H spike in the input and notch amp circuits. Let's see, I'm 74 now, so this should be finished by... um... er... heh.
Ah, I think I'm beginning to see I should
write a song:
I can see clearly now...
Well, at least on good days.
Demian, got it me thinks.
RichEEM, David,
Then here are THE "functional" mods for the HP339a.
Which gives it more flexibility.
David came up with the mod for the BNC jack on the front
panel to bypass the noisy meter section of the HP339a.
Now here is the kicker.
Put a Switch for the filters, Pre- Post- notch on a switch.
Either remove the external ground connection on the face below the AM detector.
OR
Modify the AM/GND/Lift switch to also have an UP/Down switch
for Pre-filters or Post-filters. No problem for assy work with noisy
products or quiet work on SORTA audio gear.
A WIN WIN WIN.
Doesn't get much better than that.
Puns are fun.
Good for your RichEEM we'll give you plenty to think about.
Cheers,
write a song:
I can see clearly now...
Well, at least on good days.
Demian, got it me thinks.
RichEEM, David,
Then here are THE "functional" mods for the HP339a.
Which gives it more flexibility.
David came up with the mod for the BNC jack on the front
panel to bypass the noisy meter section of the HP339a.
Now here is the kicker.
Put a Switch for the filters, Pre- Post- notch on a switch.
Either remove the external ground connection on the face below the AM detector.
OR
Modify the AM/GND/Lift switch to also have an UP/Down switch
for Pre-filters or Post-filters. No problem for assy work with noisy
products or quiet work on SORTA audio gear.
A WIN WIN WIN.
Doesn't get much better than that.
Puns are fun.
Good for your RichEEM we'll give you plenty to think about.
Cheers,
This is very good point from Demian.
The 339A has a lot of high frequency noise. it could cause problems moving the filters.
Try before hard wiring the switches in. A gain of 100 is lot in the presents of noise.
The 339A has a lot of high frequency noise. it could cause problems moving the filters.
Try before hard wiring the switches in. A gain of 100 is lot in the presents of noise.
Not picking a fight here, but my 339 doesn't have a lot of HF noise -- if I disable all filters in the measurements I've done, the meter reading just doesn't increase that much, so the 100X amp is not having to struggle with excessive noise. That said, i don't usually look at sources that are extremely noisy, but then that's what the range knob is for, and in any case, the 339 notch filter is *always* ahead of the filters, monitor output, and RMS converter.
@Sync -- the monitor output comes directly off of the A2U4 100X gain amp -- if plugging anything into the monitor jack cause big problems, then it is likely the U4 amp is going goofy -- I would think about using a different chip for that amp, like an OPA1641 -- you want a closed loop bandwidth of at least 500kHz with that stage with its 40dB of gain, and I would look to get more than 500kHz BW, and choose the feedback C accordingly -- this may require using an LT1468 and leave the 100 ohm output resistor in place.
@Sync -- the monitor output comes directly off of the A2U4 100X gain amp -- if plugging anything into the monitor jack cause big problems, then it is likely the U4 amp is going goofy -- I would think about using a different chip for that amp, like an OPA1641 -- you want a closed loop bandwidth of at least 500kHz with that stage with its 40dB of gain, and I would look to get more than 500kHz BW, and choose the feedback C accordingly -- this may require using an LT1468 and leave the 100 ohm output resistor in place.
RichEEM,
No problem. You will be pleasantly surprised.
If you don't need the filters TP5 is right next to
the connector P200 by the front panel.
It is more of a pain to down there to get to,
to jumper around it. If they aren't being used
they might not be a problem.
OPA1641 is only in SOIC currently, a break out
board for me. Using the LT1648 I'll just clip off
pin 8, but wondering how many of the other compensation
parts to keep in the circuit?
I don't know ff C25, C13, CR4, R28 and R29 should be
removed and just keep the feedback cap C11, and R12, 5K and R11 56.51.
I've got more LT1468s, I could just use it here and replace the back
to back tants of C16, C17 with electrolytics or B.P.s. Groner says
even using back to back standard electrolytics and BP electrolytics
are about the same.
Maybe changing the back to back will start to help with the meter.
I've just got a lot to check with this.
Thanks for your suggestions.
I'll post a link here RichEEM to a few picks
I took at the Klipsch Museum. It was closed
when I was there, so I just took a some outside pics.
No problem. You will be pleasantly surprised.
If you don't need the filters TP5 is right next to
the connector P200 by the front panel.
It is more of a pain to down there to get to,
to jumper around it. If they aren't being used
they might not be a problem.
OPA1641 is only in SOIC currently, a break out
board for me. Using the LT1648 I'll just clip off
pin 8, but wondering how many of the other compensation
parts to keep in the circuit?
I don't know ff C25, C13, CR4, R28 and R29 should be
removed and just keep the feedback cap C11, and R12, 5K and R11 56.51.
I've got more LT1468s, I could just use it here and replace the back
to back tants of C16, C17 with electrolytics or B.P.s. Groner says
even using back to back standard electrolytics and BP electrolytics
are about the same.
Maybe changing the back to back will start to help with the meter.
I've just got a lot to check with this.
Thanks for your suggestions.
I'll post a link here RichEEM to a few picks
I took at the Klipsch Museum. It was closed
when I was there, so I just took a some outside pics.
Last edited:
@Sync -- thx for the pix -- I haven't been back to Hope since the late 70s, when I lived in Little Rock. Paul asked me to come back down for a consult on tweaking his log amp rack.
RE U4 -- remove C13, C25, CR4, R28 and R29 -- then you don't have to clip pin 8 of the 1468. I would change C11 to 22pF. Leave C15 in for a HF stopper.
RE U4 -- remove C13, C25, CR4, R28 and R29 -- then you don't have to clip pin 8 of the 1468. I would change C11 to 22pF. Leave C15 in for a HF stopper.
A 11MHz op amp (1641) would only give you 110kHz at a gain 100. I used the LT1468 for U4. Works very well. You get the full 90MHz GBW at G=100. That will get you your 500kHz BW.
Last edited:
Gentlemen,
Got it. Almost had the pull out A4 board redone but
well you know, life happens and my little girl wants
to eat dinner, play, needs diaper change, etc.
She's a pretty good kid.
I can't figure out why she cry's
when I give her a refridgerator
magnet ice cream cone though...
"Aye Scream
You Scream
We all Scream
For Ice Cream."
Then she cry's
when she tries to
eat it. -- shame on me.
Even worse,
she has a choice
Chocolate or Strawberry.
Cheers,
Forget bier I need ALE!
Got it. Almost had the pull out A4 board redone but
well you know, life happens and my little girl wants
to eat dinner, play, needs diaper change, etc.
She's a pretty good kid.
I can't figure out why she cry's
when I give her a refridgerator
magnet ice cream cone though...
"Aye Scream
You Scream
We all Scream
For Ice Cream."
Then she cry's
when she tries to
eat it. -- shame on me.
Even worse,
she has a choice
Chocolate or Strawberry.
Cheers,
Forget bier I need ALE!
Last edited:
RichEEM,
No problem. You will be pleasantly surprised.
If you don't need the filters TP5 is right next to
the connector P200 by the front panel.
It is more of a pain to down there to get to,
to jumper around it. If they aren't being used
they might not be a problem.
OPA1641 is only in SOIC currently, a break out
board for me. Using the LT1648 I'll just clip off
pin 8, but wondering how many of the other compensation
parts to keep in the circuit?
I don't know ff C25, C13, CR4, R28 and R29 should be
removed and just keep the feedback cap C11, and R12, 5K and R11 56.51.
I've got more LT1468s, I could just use it here and replace the back
to back tants of C16, C17 with electrolytics or B.P.s. Groner says
even using back to back standard electrolytics and BP electrolytics
are about the same.
Maybe changing the back to back will start to help with the meter.
I've just got a lot to check with this.
Thanks for your suggestions.
I'll post a link here RichEEM to a few picks
I took at the Klipsch Museum. It was closed
when I was there, so I just took a some outside pics.
Sync the LT1468 data sheet has a formula for calculating the FB cap. And there is more in a paper called LT1468 design features which may give more insight in how to use the 1468. The FB cap provides necessary current to balance the input capacitance of the 1468 + whatever stray capacitance is encountered at it's input. The input capacitance causes high frequency peaking in the closed loop gain. As the FB resistance increases the FB cap value must increase proportionally with it. Sometimes it's best to find the right FB cap value empirically. Always use the smallest cap that does the job.
David,
To find the smallest value feedback cap, I look for oscillations, distortion,
clipping on the output with scope, raise feedback cap value until it stops.
Provided the feedback resistor isn't out of line.
I have the LT1468 data sheet and application notes.
I was trying to figure out how to reconcile the differences
between the data sheet and application note with the circuits.
In this case the different circuits in the HP339a and where
the tradeoffs are for current noise and voltage noise.
I know now there is a difference between FET Opamps
and Bipolar Opamps. I can't tell which circuit needs what
and why? I know you and others have mentioned it a few
times.
Every now an again the light does come on.
Even if it is only a C7 or a C9 when I want a
200 Watt flood.
Cheers,
PS - I eventually saved the day and broke
out the Neapolitan Ice Cream out of the freezer.
Her joy put a great big smile upon my face.
It is these times times I just love being DAD !
Thank you each and every one during this holiday season.
To find the smallest value feedback cap, I look for oscillations, distortion,
clipping on the output with scope, raise feedback cap value until it stops.
Provided the feedback resistor isn't out of line.
I have the LT1468 data sheet and application notes.
I was trying to figure out how to reconcile the differences
between the data sheet and application note with the circuits.
In this case the different circuits in the HP339a and where
the tradeoffs are for current noise and voltage noise.
I know now there is a difference between FET Opamps
and Bipolar Opamps. I can't tell which circuit needs what
and why? I know you and others have mentioned it a few
times.
Every now an again the light does come on.
Even if it is only a C7 or a C9 when I want a
200 Watt flood.
Cheers,
PS - I eventually saved the day and broke
out the Neapolitan Ice Cream out of the freezer.
Her joy put a great big smile upon my face.
It is these times times I just love being DAD !
Thank you each and every one during this holiday season.
Last edited:
You can also run the circuit in spice with a loop gain probe. But this won't include strays.
It will get you into the ball park.
For a 5k FB resistor the 10pF HP has in there will probably do the job.
I usually find 7pF to 10pF is about right.
What matters here is the GBW. Not so much fet or bipolar. The input is seeing a low Z from the op amp driving it. The input R of the LT1468 is 150k !! 1M and 0.22uF coupling cap puts the HP at 5.55Hz which in low enough. With the 2625's 300M ohm input R it was about 0.73Hz.
A fet input op amp would be the 2pieRC of the coupling cap and 1M ohm resistor at the input of U4 since a Jfet's gate resistance is negligible. A jfet op amp would work here if we can find one with wide enough bandwidth. The current noise is so low you don't even need to look at it. The LT1468 is a perfect compromise between bipolar and jfet input op amps. it certainly beats most for distortion.
An ultra low noise opa1611 for example has an input R around 20k. This is typical so it wouldn't work to use one of the ultra low noise parts here. Besides they all seem to have relatively poor current noise specs. They have to be used with low Z FB and input R's to benefit.
HP had a reason for doing it the way they did. So we don't want to mess with it too much.
RE your daughter, enjoy it while she is young. Move out when she's a teenager.
or at least build shed in the back yard.
Last edited: