Date
1 - 13 of 13
About Polyphase network in receiver
Leif M
A receiver with a Tayloe detector seems to be a good way to listen to HW. For SSB I need polyphase network, that much I know.
Polyphase network is critical. But there are versions 4 inputs and others with I and Q inputs. A book called Experimental methods in RF design gives some formulas. But it is a bit difficult to find instructions which fit together, and give part values. It looks like I should buy the whole thing.
But the polyphase network I would make differently myself. I would use parts I linked below.
There are capacitors with +-1% tolerance and SMD resistors with 0.01% tolerance at Digikey, and surely elsewhere too.
Here is one 22nF capacitor
https://www.digikey.com/en/products/detail/vishay-roederstein/MKP1837322161G/5392952
And a 0.01% resistor for example. These are 0805
https://www.digikey.fi/en/products/detail/susumu/URG1608L-472-L-T05/4552652
These resistors are not cheap, but there are others less accurate, like 0.1%. They are probably good enough.
If I buy a receiver with a Tayloe detector, can I or should I use these components to listen to SSB? What if I design a PCB with part I showed.
Polyphase network is critical. But there are versions 4 inputs and others with I and Q inputs. A book called Experimental methods in RF design gives some formulas. But it is a bit difficult to find instructions which fit together, and give part values. It looks like I should buy the whole thing.
But the polyphase network I would make differently myself. I would use parts I linked below.
There are capacitors with +-1% tolerance and SMD resistors with 0.01% tolerance at Digikey, and surely elsewhere too.
Here is one 22nF capacitor
https://www.digikey.com/en/products/detail/vishay-roederstein/MKP1837322161G/5392952
And a 0.01% resistor for example. These are 0805
https://www.digikey.fi/en/products/detail/susumu/URG1608L-472-L-T05/4552652
These resistors are not cheap, but there are others less accurate, like 0.1%. They are probably good enough.
If I buy a receiver with a Tayloe detector, can I or should I use these components to listen to SSB? What if I design a PCB with part I showed.
Alan G4ZFQ
On 19/03/2023 21:20, Leif M wrote:
I see links here http://qrp-labs.com/polyphase.html that seem to say extreme accuracy is not so important if attention is made to good matching and component selection.
See the pdf linked there.
73 Alan G4ZFQ
There are capacitors with +-1% tolerance and SMD resistors with 0.01% tolerance at Digikey, and surely elsewhere too.Leif,
I see links here http://qrp-labs.com/polyphase.html that seem to say extreme accuracy is not so important if attention is made to good matching and component selection.
See the pdf linked there.
73 Alan G4ZFQ
ajparent1/kb1gmx
You do not have to measure them. Its far easier to MATCH them.
Also QRPL supplies a Polyphase network to match their receiver.
See...
http://qrp-labs.com/receiver.html
http://qrp-labs.com/polyphase.html
from those pages find information on component matching.
--
Allison
------------------
Post online only, please no email.
Also QRPL supplies a Polyphase network to match their receiver.
See...
http://qrp-labs.com/receiver.html
http://qrp-labs.com/polyphase.html
from those pages find information on component matching.
--
Allison
------------------
Post online only, please no email.
Alan G4ZFQ
On 20/03/2023 17:00, Leif M wrote:
I made this https://sites.google.com/site/vk3bhr/home/lcm1
and tested this https://sites.google.com/site/g4zfqradio/lc100a-terminals-on-the-right
While testing requires some close tolerance capacitors you are not looking for accuracy, just a close match.
Is 1% actually required? Unless you have very good test techniques you will not know the difference.
73 Alan G4ZFQ
where or how I can measure capacitors better than at +-1% accuracyLeif,
I made this https://sites.google.com/site/vk3bhr/home/lcm1
and tested this https://sites.google.com/site/g4zfqradio/lc100a-terminals-on-the-right
While testing requires some close tolerance capacitors you are not looking for accuracy, just a close match.
Is 1% actually required? Unless you have very good test techniques you will not know the difference.
73 Alan G4ZFQ
Pim Niessen
Leif,
- Calculating values for a Polyphase network can be done in Spice or with this spreadsheet: http://antennoloog.nl/data/documents/Polyphase_PA2PIM_2.4-20060324.zip
- The difference is that LT-Spice takes into account all kind of parasitic parameters of the components. The spreadsheet does not. But I'm not sure how much influence that has, probably not much in the 200-4000 Hz range.
- Also LT-Spice has a bit of a learning curve which is much less with the spreadsheet.
- Matching components values within one section is way more important then the absolute value of components.
- Poor matching reduces the unwanted sideband suppression. A shift in the absolute value of the components within a section will shift the frequency where that section produces the best suppression.
- The measuring device mentioned above, and variants of it, are a worthwhile piece of kit. The more so with SMD components with no or undecipherable markings on them.
- But since only matching is important, this circuit from 1990 also works fine.
- There are some more hints on how to squeeze the best performance out of a bag of components, see paragraph 8 in: http://antennoloog.nl/data/documents/Understanding_and_designing_Polyphase_networks_V4.0.pdf
- You mention purchasing a receiver with a Tayloe detector and polyphase network.
- The one from QRP lab is optimized for 700 -1500 Hz audio as Hans states, see also the Blue graph here: http://qrp-labs.com/polyphase.html
- If you want to have better unwanted sideband suppression over the SSB audio frequency range you will need to add sections to the Polyphase network.
- The only other source of kits with Polyphase networks I know of are the Juma series by OH7SV and friends. Not sure if they are still available but price range was quite different from QRP-lab.
John Kirby
And here is very nice (free) PC Sound Card Oscilloscope download with spectrum analyzer and function generator
>>>https://www.zeitnitz.eu/scope_en<<<
If components to build Pim's PA2PIM excellent suggestion are not available...
...a smile "parallel" L/C network plus a 'grid dip' meter or SWR Meter work well, you supply the C then measure frequency then cull the "best match"
72 73
John
N3AAZ
>>>https://www.zeitnitz.eu/scope_en<<<
If components to build Pim's PA2PIM excellent suggestion are not available...
...a smile "parallel" L/C network plus a 'grid dip' meter or SWR Meter work well, you supply the C then measure frequency then cull the "best match"
72 73
John
N3AAZ
Johan Bodin
Many thanks Pim,
the best I have ever read about polyphase filter design!
73 Johan SM6LKM
Pim Niessen wrote:
the best I have ever read about polyphase filter design!
73 Johan SM6LKM
Pim Niessen wrote:
http://antennoloog.nl/data/documents/Polyphase_PA2PIM_2.4-20060324.zipand
http://antennoloog.nl/data/documents/Understanding_and_designing_Polyphase_networks_V4.0.pdf
vbifyz
My experience is not about polyphase, but the all-pass opamp circuit. I think it is still somewhat relevant.
I matched 10nF capacitors from a bag of 50 5% parts to be within 1% of each other, using a $25 LC-meter/transistor tester from eBay.
Resistors are 1% from E24 row (so the values are not ideal), no matching.
LTSpice shows about 40dB unwanted SSB suppression over 300..3000Hz range, the real life value is very close.
With other RX circuits added, the suppression is a bit worse, about 35dB, because of the phase mismatch in RF components. So the precision on the AF side is actually not that important.
This is still good enough for a homebrew project.
73, Mike AF7KR
I matched 10nF capacitors from a bag of 50 5% parts to be within 1% of each other, using a $25 LC-meter/transistor tester from eBay.
Resistors are 1% from E24 row (so the values are not ideal), no matching.
LTSpice shows about 40dB unwanted SSB suppression over 300..3000Hz range, the real life value is very close.
With other RX circuits added, the suppression is a bit worse, about 35dB, because of the phase mismatch in RF components. So the precision on the AF side is actually not that important.
This is still good enough for a homebrew project.
73, Mike AF7KR
Johan Bodin
I have built all-pass phase shifters a few times and they do work well when carefully adjusted. QuadNet software is a convenient design tool http://tonnesoftware.com/quad.html.
73 Johan SM6LKM
vbifyz wrote:
toggle quoted message
Show quoted text
73 Johan SM6LKM
vbifyz wrote:
My experience is not about polyphase, but the all-pass opamp circuit. I think it is still somewhat relevant.
I matched 10nF capacitors from a bag of 50 5% parts to be within 1% of each other, using a $25 LC-meter/transistor tester from eBay.
Resistors are 1% from E24 row (so the values are not ideal), no matching.
LTSpice shows about 40dB unwanted SSB suppression over 300..3000Hz range, the real life value is very close.
With other RX circuits added, the suppression is a bit worse, about 35dB, because of the phase mismatch in RF components. So the precision on the AF side is actually not that important.
This is still good enough for a homebrew project.
73, Mike AF7KR
Pim Niessen
Johan, e.a.
After glancing at the tonnesoftware I realized that LTSpice can also do Monte Carlo Analysis.
Mr. Gingell, the inventor of the Polyphase networks, states that he can achieve 60 dB of unwanted sideband suppression when he grades the tolerances from in- to output from 2,5 to 0.2 %
See last page of: https://rubidium.se/~magnus/synths/friends/gingell/1973itt.pdf
It will be interesting to do the same with an all-pass circuit, compare the outcomes, and finally nail the discussion on influence of component accuracy and stability between the two types of networks.
Interesting pass-time for a rainy afternoon.
Pim
Hans Summers
Hi Pim
As I'm sure you know but for the benefit of everyone else - the polyphase network is said to have greater unwanted sideband suppression compared to all-phase networks using components with the same tolerance.
However there are also other factors involved. A criticism of polyphase networks is that they do not have a flat response across the audio passband. This is in fact not necessarily true; if one abandons having all the capacitors the same value (or resistors the same value) and is prepared to change both resistor values AND capacitor values from column to column, then the voltage gain of the network can be set to 1.0 and very flat across the audio passband.
Yet another factor is the noise figure of a polyphase network - given that it is composed of resistive elements as well as capacitive... op-amps of course also have an input noise. The question becomes more complex than simply unwanted sideband cancellation since now you have to include the rest of the receiver design in the question, op-amp performance, system gain distribution and impedances.
A lot of ins and outs. I have used polyphase networks in homebrew rigs and in the QRP Labs receiver module, and active all-pass networks in QCX. So some experience of both, and could say that I have always been pleased with the results, both ways :-)
On Wed, Mar 22, 2023 at 4:35 PM Pim Niessen <pa2pim@...> wrote:
Johan, e.a.
After glancing at the tonnesoftware I realized that LTSpice can also do Monte Carlo Analysis.
Mr. Gingell, the inventor of the Polyphase networks, states that he can achieve 60 dB of unwanted sideband suppression when he grades the tolerances from in- to output from 2,5 to 0.2 %
See last page of: https://rubidium.se/~magnus/synths/friends/gingell/1973itt.pdf
It will be interesting to do the same with an all-pass circuit, compare the outcomes, and finally nail the discussion on influence of component accuracy and stability between the two types of networks.
Interesting pass-time for a rainy afternoon.
Pim