UK Vintage Radio Repair and Restoration Powered By Google Custom Search Vintage Radio and TV Service Data

Go Back   UK Vintage Radio Repair and Restoration Discussion Forum > Other Discussions > Homebrew Equipment

Notices

Homebrew Equipment A place to show, design and discuss the weird and wonderful electronic creations from the hands of individual members.

Closed Thread
 
Thread Tools
Old 2nd Jun 2021, 8:27 pm   #21
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: RF preamp circuit selection

I recall experimenting with it to look for possible improvements in unwanted cross-modulation or other distortion artefacts but I got bad results for distortion with or without it.
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline  
Old 2nd Jun 2021, 8:35 pm   #22
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,800
Default Re: RF preamp circuit selection

Alternatively, do the Y factor measurement with the intended IF amplification etc in place and you get the system NF with low uncertainty and no need for Mr Friis.

David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done
Radio Wrangler is offline  
Old 2nd Jun 2021, 8:50 pm   #23
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: RF preamp circuit selection

I wondered about that but I got in a hopeless muddle worrying about the effect of the nearby image noise generated from the noise source during the first part of the measurement.

Going back to the clunky sig gen method I tried a comparison with and without the SMD tant cap, i.e. I tried no centre tap capacitor at all with the 10kHz IF.

The conversion gain is identical in both cases but the noise level either side of 10kHz jumps up 6dB on the analyser when the centre tap tantalum cap is removed. That is quite a difference! I doubt this would show any effect up at a 10.7MHz IF but I can try this later.

I now don't think I need to make any calibrated or exotic measurements because I can just look at the analyser and see the noise leap up 6dB when the centre tap cap is pinged off. I didn't expect to see this much degradation.
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline  
Old 2nd Jun 2021, 9:10 pm   #24
Alan_G3XAQ
Pentode
 
Join Date: Apr 2018
Location: Canterbury, Kent, UK.
Posts: 189
Default Re: RF preamp circuit selection

Hmm. I've made the audio stage a Q=5 BPF. No change in noise floor. Still -120dBm. I guess it's time to start looking for ground loops and/or improve the shielding between the LO and log detector?

73, Alan
Alan_G3XAQ is offline  
Old 2nd Jun 2021, 9:18 pm   #25
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: RF preamp circuit selection

I can try this on one of the other two eval boards or I could just swap out the mixer for another one. I still have quite a few of the SA/NE/602A/612A in DIL and SMD flavours in IC tubes here but they are old stock from the 1990s.

I should also point out that I'm using an SMD SA602A and it will be at least 25 years old. I'm also using an external LO (Marconi 2024 sig gen) rather than a crystal based LO.
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline  
Old 2nd Jun 2021, 9:26 pm   #26
Alan_G3XAQ
Pentode
 
Join Date: Apr 2018
Location: Canterbury, Kent, UK.
Posts: 189
Default Re: RF preamp circuit selection

Do you have a theory as to why the noise jumps up 6dB without the AC centre tap on the input? Are the internal bias circuits noisy at 10KHz?
Alan_G3XAQ is offline  
Old 2nd Jun 2021, 9:33 pm   #27
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: RF preamp circuit selection

Quote:
Originally Posted by Alan_G3XAQ View Post
Hmm. I've made the audio stage a Q=5 BPF. No change in noise floor. Still -120dBm. I guess it's time to start looking for ground loops and/or improve the shielding between the LO and log detector?

73, Alan
I think you have to rule out these possibilities because they may dominate the noise compared to the SA602 even with poor LF decoupling at the RF inputs.

Also, there needs to be a useful amount of gain ahead of the 10kHz BPF to make up a valid noise pedestal at 10kHz. Otherwise you end up with a BPF that filters nicely against external (antenna) signals but it won't reduce the effective noise bandwidth arriving at the logamp.
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline  
Old 2nd Jun 2021, 9:37 pm   #28
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: RF preamp circuit selection

Quote:
Originally Posted by Alan_G3XAQ View Post
Do you have a theory as to why the noise jumps up 6dB without the AC centre tap on the input? Are the internal bias circuits noisy at 10KHz?
It seems that the bias circuits are quite noisy and Jim mentioned this near the start of the thread. I had a quick look on a wider span and this problem extends out to hundreds of kHz but the added noise does gradually get lower as the frequency goes up. I might have a duff mixer as this one has been used a lot. I have three eval boards and plenty of these mixers here!
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline  
Old 2nd Jun 2021, 9:55 pm   #29
Alan_G3XAQ
Pentode
 
Join Date: Apr 2018
Location: Canterbury, Kent, UK.
Posts: 189
Default Re: RF preamp circuit selection

Quote:
Originally Posted by G0HZU_JMR View Post
Quote:
Originally Posted by Alan_G3XAQ View Post
Hmm. I've made the audio stage a Q=5 BPF. No change in noise floor. Still -120dBm. I guess it's time to start looking for ground loops and/or improve the shielding between the LO and log detector?

73, Alan
I think you have to rule out these possibilities because they may dominate the noise compared to the SA602 even with poor LF decoupling at the RF inputs.

Also, there needs to be a useful amount of gain ahead of the 10kHz BPF to make up a valid noise pedestal at 10kHz. Otherwise you end up with a BPF that filters nicely against external (antenna) signals but it won't reduce the effective noise bandwidth arriving at the logamp.
OK, so I will add some shielding and see if it makes any difference. I have tried injecting a signal after the preselector xtal filter and confirmed the bandwidth of the audio is about 2KHz, so the BPF is working.

I'll rewind the mixer input transformer with a centre tap tomorrow. I don't see a need for a fancy tantalum cap, which I don't have. Modern garden variety Aluminium electrolytics should be OK at 10KHz?

I'm unclear about the noise pedestal you talk about. I'm very short on physical space around the 10KHz stage so adding extra gain there will be problematic. Are there any other options?

73, Alan
Alan_G3XAQ is offline  
Old 2nd Jun 2021, 10:17 pm   #30
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: RF preamp circuit selection

You should be able to repeat the tests I'm doing if you still have your HP8568B analyser. I think the noise floor below 100kHz on the DC coupled input is much lower than the figure given in the datasheet.

I can't remember for sure but my HP8568B always had a noise floor of -150dBm/Hz with 0dB attenuation across HF and into VHF. At lower frequencies this does creep up but I'm pretty sure mine still managed something like -140dBm/Hz in the lowish kHz region and close to -130dBm/Hz with the default internal 10dB attenuator.

I can show you the BJT lashup interface I used to interface the IF port of the SA602A to the 50R analyser? I'm afraid discrete design at AF isn't my thing but the BC107 datasheet shows noise circles with a sub 2dB noise figure with a source impedance of 1500R at 10kHz and a collector current of 1mA.

I put together a DC coupled amp using a pair of BC547B BJTs and the emitter of the second stage has enough grunt to drive the analyser via a series 56R resistor. I'm seeing a noise floor of -119dBm/Hz at 10kHz on the analyser due to the SA202 mixer and the BJT AF amplifier stage.
If I disconnect the SA602 and bias the input of the BJT from 4V via 1500R then the noise floor drops a lot. So the -119dBm/Hz seen on the analyser is dominated by the SA602 with this setup. When I remove the tap capacitor at the input transformer of the SA602A this noise level on the analyser jumps up 6dB so this is fairly damning I think.

The circuit is below and an audio guru could probably design something a lot better than this! The 4V source SG2 and the 1500R bias resistor R6 shown at the input represents what the datasheet implies is at pins 4 and 5 of the SA602A so I just make a direct connection of the base of the BC547B to pin 4 of the SA602A and the 4V bias and R6 is provided for fee by the innards of the SA602A.

Obviously, this temporary BJT interface is very puny and only suitable for small signals but I'm just measuring close to the noise floor here.

Your HP8568B noise floor should be much lower than this -119dBm/Hz even with the default 10dB attenuator. You just have to make sure you don't let any significant DC current get into the DC coupled input port of the analyser or you could toast the analyser front end. The 12V supply would need to be current limited to just a few mA in case something went wrong with the connections.
Attached Thumbnails
Click image for larger version

Name:	AF_DC_AMP_602A.jpg
Views:	73
Size:	29.7 KB
ID:	235284  
__________________
Regards, Jeremy G0HZU

Last edited by G0HZU_JMR; 2nd Jun 2021 at 10:27 pm.
G0HZU_JMR is offline  
Old 2nd Jun 2021, 10:19 pm   #31
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,800
Default Re: RF preamp circuit selection

No problem, Jeremy. The Y factor measurement would be a DSB frequency conversion one, but then , so is the measurement of environmental noise the thing is being made to do, so it all comes out in the wash.

Back in 2000, the HP noise figure applications notes were re-written (they may have been redone since to mark the end of the noise figure analyser and to steer people to apps in spectrum analysers) But the 2000 editions were not throttled back. AN57-1 was written by Bruce Erickson and gives the general overview. AN57-2 goes into decent detail and explains the maths and the sources of error (and where the bodies are buried) writing it was contracted out because everyone was too busy. It first went out to a university, where a prof handed it out to an inexperienced student. ~The result was not useable, but most of the time got used up! We were rescued by Ian White, G3SEK. Ian was working as a professional technical author with clients like British Nuclear Fuels, Shell Expro and others in that league. His interest in low noise RF was perfect. I suggested Ian to the marketing folk knowing he already had the background knowledge and would take to it like the proverbial duck. Well, those who know the field were very pleased with the result.

Duncan Boyd contributed his web-based uncertainty calculator.
I'd adjusted the definition of the definition of ENR. Before, it was only valid at one spot temperature. I added the appropriate temperature term to the noise contributed by the resistive termination/attenuator in a noise source so that at least that variance was properly tracked and compensated. I'd added a digital thermometer in the new noise sources.

David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done
Radio Wrangler is offline  
Old 2nd Jun 2021, 10:28 pm   #32
Alan_G3XAQ
Pentode
 
Join Date: Apr 2018
Location: Canterbury, Kent, UK.
Posts: 189
Default Re: RF preamp circuit selection

I can sort-of do the equivalent test directly: disconnect the mixer output from the audio input. I will do this tomorrow when I'm more awake.

I did an almost-equivalent test earlier by disconnecting the DC power to the mixer. The log detector output dropped 14dB. This was with a broadband audio stage. Yes, removing power from the mixer stops the LO so the test didn't rule out LO leakage.

I still don't have a grasp on your noise pedastal comments and the need for added gain.

73, Alan
Alan_G3XAQ is offline  
Old 2nd Jun 2021, 10:46 pm   #33
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: RF preamp circuit selection

You might not need to worry about the noise bandwidth because your IF section might already be OK in this respect. However, here's a plot of a typical noise pedestal after a 9MHz crystal filter.

When the signal is turned on and off to do a sensitivity test the noise power that contributes to S+N/N is dominated by the 3kHz wide pedestal of noise and not the mean power contained in any noise floor either side of it that might be much wider in bandwidth.

To get a raised pedestal of noise like this there has to be some gain ahead of the bandwidth defining filter. In a regular receiver this is less important because our ears act as a bandwidth limiting filter. But a wideband limiter or logamp will just allow everything in it can detect over a huge bandwidth and it will add it to the S+N/N equation when the signal is turned off.
Attached Thumbnails
Click image for larger version

Name:	NoisePedestal.gif
Views:	48
Size:	24.8 KB
ID:	235286  
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline  
Old 2nd Jun 2021, 11:20 pm   #34
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: RF preamp circuit selection

Quote:
I'll rewind the mixer input transformer with a centre tap tomorrow. I don't see a need for a fancy tantalum cap, which I don't have. Modern garden variety Aluminium electrolytics should be OK at 10KHz?
For this test it probably will be OK as it is at the centre tap of the winding so it is unlikely to interfere with the RF performance and it is bound to do something at 10kHz. I did read stuff about the merits of using low ESR tantalum caps for the SA602A in an old Philips app note but I think this was for the usual supply decoupling/filtering. Try it!
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline  
Old 3rd Jun 2021, 5:05 pm   #35
Alan_G3XAQ
Pentode
 
Join Date: Apr 2018
Location: Canterbury, Kent, UK.
Posts: 189
Default Re: RF preamp circuit selection

I rewound the mixer input transformer to add a centre tap and AC coupled the tap to ground with 1uF plus 10nF capacitors. I also made some more careful measurements of the return loss seen by the link winding and increased the turns from 2 to 5 (28t with a CT, tuned with a 65pF film trimmer), suggesting the coefficient of magnetic coupling is around 0.6.

These changes have improved the MDS by 10dB. I can now clearly see a -130dBm carrier on the output meter. The log detector output is close to the datasheet noise floor (400mV) and hardly changes when the mixer is on or off. I could probably use more gain but I don't have space for more amplifiers so it will have to suffice as it is.

With a 1.3V fsd meter and close to 400mV zero offset from a forward biased Schottky diode I see the meter rise a few percent above zero with -127dBm carrier (= S0) and get very close to fsd with -73dBm (=S9). This is close to ideal for my intended use so I will screw the lid on the box now and do something else.

Thanks everyone,

Alan G3XAQ
Alan_G3XAQ is offline  
Old 3rd Jun 2021, 8:04 pm   #36
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,800
Default Re: RF preamp circuit selection

In a normal receiver design there is progressive gain, so that signal levels increase as they flow down the structure. There are different bandwidths along the way and trying to get the best dynamic range tries to push the narrowest bandwidtm the channel defining filter as early as possible.

The result is a set of noise pedestals, maybe just one or maybe a mini-ziggurat!

In Jeremy's photo you can see the broadband noise pedestal set by the noisiness of the post-crystal-filter stages and their bandwidth being wider than the crystal filter. Through the bandwidth of the crystal filter you see the noise floor of your input stages boosted by their gain.

You won't see any bandwidth effects from selectivity before the crystal filter because its bandwidth is dominant.

So with a receiver operating in an environment with noise and many signals plus also different noise from its various stages, it really is a serious game to work out how much of what reaches the detector and at what levels.

Remember also that phase noise on LOs and also on ADC clocks can also give noise-pedestal shapes which through reciprocal mixing can appear to transfer onto each and all signals. passing through the mixer with that LO. The pedestal on a large signal can swamp the whole of a small signal.

More exotic is the business of broadband noise accompanying the signals convolving with broadband noise accompanying the LO. This can lead to a surprising amount of noise getting into your IF. Sometimes called the noise-noise convolution problem.

I ran into the latter a lot with NPR (noise power ratio) testing of the linearity of FDM type multichannel phone links (2600 channels) where 60kHz to 18.6MHz of white noise was used to simulate a fully loaded phone system. If there was broadband noise from untuned LO amps we quickly got into noisexnoise problems.

David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done
Radio Wrangler is offline  
Old 3rd Jun 2021, 10:03 pm   #37
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: RF preamp circuit selection

A good result! When Jim mentioned the bias noise issue earlier in the thread I didn't realise it was this significant for the SA602 down at a very low IF.

I just powered up the other eval board and this has a 50MHz single ended input transformer and a 10.7MHz IF. The IF BW is quite broad as I use an L match to match 1500R to 50R at the IF port.

This only has 100nF decoupling to ground the single ended input transformer.
I measured it using the analyser and noise source and it showed a nice smooth noise figure response that was at its best at 50MHz and this showed a 4.7dB noise figure. The conversion gain was just over 14dB. Therefore, the 100nF decoupling is adequate with an IF up at 10.7MHz.

From memory the other eval board showed something like 4.9dB noise figure and just over 17dB conversion gain with the balanced input transformer with the same 10.7MHz IF.

I'm going to try and measure the IF noise response at the mixer pins 4 and 5 with and without the xfmr centre tap (tant) capacitor in place and I'll do this from a few kHz out to a few MHz if possible. This should hopefully show where the extra decoupling becomes necessary as the IF frequency is reduced.
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline  
Old 3rd Jun 2021, 10:52 pm   #38
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: RF preamp circuit selection

Quote:
No problem, Jeremy. The Y factor measurement would be a DSB frequency conversion one, but then , so is the measurement of environmental noise the thing is being made to do, so it all comes out in the wash.
Thanks. I'm going to try a few experiments. At regular RF frequencies the PSA analyser can select LSB, USB or DSB when it can support both the RF and IF frequencies for noise measurements in the frequency converter mode. In the case of the 50MHz input with 61MHz LO and 11MHz IF I select LSB, DOWNCONVERTER as in the image below.
I've only used the converter mode a few times when measuring noise figure but it is a bit fiddly to set up to make sure it calibrates correctly with the correct ENR values for both IF and RF calibration. The analyser GUI does have an onscreen wizard that helps with the setup and there's screenshot below. I guess you have seen this GUI a few times over the years.

However, when I try do the uncorrected measurement with the 10kHz IF the analyser doesn't support an IF this low so I plan to use the noise source for the 50MHz input and the Tek RTSA analyser for the IF port measurement as it has a very low noise floor down at a few kHz. It is typically -155dBm/Hz.
I'm just going to measure hot and cold noise levels and then try and make sense of the result because of the DSB mode of operation.
Attached Thumbnails
Click image for larger version

Name:	NF_setup.gif
Views:	40
Size:	13.1 KB
ID:	235395  
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline  
Old 4th Jun 2021, 5:17 pm   #39
G0HZU_JMR
Dekatron
 
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
Default Re: RF preamp circuit selection

I had some more free time this afternoon and I had a go at doing the noise floor comparison at the IF port via the BJT based IF stage into a spectrum analyser.

This is for a SA602A mixer being fed with a balanced transformer across pins 1 and 2 at 50MHz. The plots are of the IF noise and they compare the effect of adding a large value tantalum cap (47uF) at the centre tap of the secondary of the RF transformer.

This does show the noise reduction when the cap is added is quite significant up to a few hundred kHz. At 10kHz the noise with the centre tap decoupling cap is just over 5dB lower. The noise is still slightly lower even out at an IF of 1MHz.

Note that to rule out the possibility of my bench supply injecting wideband regulator noise into the SA602A these plots were taken with the SA602A powered from a PP3 battery with a voltage dropping RC filter adjusted to give just over 5V at the supply pin of the SA602A. Therefore, this noise really is generated by the SA602A. I also tried various signal sources for the external RF LO and the results were the same. It was the same with the PP3 battery replaced with a decent linear bench PSU.

The plots below are quite surprising to me. It obviously is important to decouple both the supply pins of the SA602A and the RF input. The Philips app notes for the SA602A recommends low ESR tant caps for the decoupling but it doesn't describe this particular noise issue when using a low frequency IF.

Both of the plots below are scaled at 1dB/div and the first plot shows the noise out to 100kHz and the second shows the noise out to 1MHz.
Attached Thumbnails
Click image for larger version

Name:	SA602A_100kHz_noise.gif
Views:	54
Size:	20.8 KB
ID:	235428   Click image for larger version

Name:	SA602A_1MHz_noise.gif
Views:	45
Size:	17.4 KB
ID:	235429  
__________________
Regards, Jeremy G0HZU
G0HZU_JMR is offline  
Old 4th Jun 2021, 5:26 pm   #40
Alan_G3XAQ
Pentode
 
Join Date: Apr 2018
Location: Canterbury, Kent, UK.
Posts: 189
Default Re: RF preamp circuit selection

It's pretty black and white, isn't it Jeremy? I have 100u+100n on the supply to my mixer but the cramped space meant I cheated and only used 1u+10n on the input centre tap. Even X=j16 seems to have made a big improvement to my receiver.

Alan
Alan_G3XAQ is offline  
Closed Thread

Thread Tools



All times are GMT +1. The time now is 12:13 am.


All information and advice on this forum is subject to the WARNING AND DISCLAIMER located at https://www.vintage-radio.net/rules.html.
Failure to heed this warning may result in death or serious injury to yourself and/or others.


Powered by vBulletin®
Copyright ©2000 - 2024, vBulletin Solutions, Inc.
Copyright ©2002 - 2023, Paul Stenning.