|
New to Ham Radio?
My Profile
Community
Articles
Forums
News
Reviews
Friends Remembered
Speak Out
Strays
Survey Question
Operating
Contesting
DX Cluster Spots
Propagation
Resources
Calendar
Classifieds
Ham Exams
Ham Links
List Archives
News Articles
Product Reviews
QSL Managers
Site Info
eHam Help (FAQ)
Support the site
The eHam Team
Advertising Info
Vision Statement
About eHam.net
|
|
RE: 7360 noise
|
Reply
|
|
by N2EY on April 15, 2009
|
Mail this to a friend!
|
Peter,
Just a quick note about the Pullen mixer.
It's a two-triode circuit that has a reputation for low noise, high conversion gain and high dynamic range. Whether that reputation is deserved I do not know.
The Pullen circuit looks like a cathode-coupled triode mixer, with the distinction that one triode has its grid leak returned to ground while the other has its grid leak returned to cathode. But there's more to it than the circuit; there's a whole design method shown in Pullen's book (see below).
Googling "pullen mixer" brought up a wealth of info, including excerpts from Pullen's book showing design methods, articles from 73 magazine comparing various tubes in the circuit, the W2VCZ version using the 6ES8 in a 75A-4, and a discussion from the AM Radio forum about the SS-1R and the Pullen mixer (including measurements of SS-1R MDS and dynamic range).
The Pullen circuit appears in the 4th edition of the RSGB Handbook but is not identified as such.
Gotta run (literally)
73 de Jim, N2EY
|
| |
|
RE: 7360 noise
|
Reply
|
|
by N2EY on April 15, 2009
|
Mail this to a friend!
|
Hello Peter,
More detailed response to your comments.
G3RZP: "By using the usual equations for pentodes and tetrodes, I can get 1500 ohms for an amplifier. But this just can’t apply to a mixer."
I think the beam-deflection mixers may be a special case because they 'mix' in a different way than, say, a pentagrid. In any event, what really matters is whether the resulting no-RF-stage receiver has adequate sensitivity for the intended application.
When it appeared, the SS-1R was in the same price class as the Collins 75S-3. I think that proved to be its undoing, because the S-line was a well-known rx from a company with a known reputation, and it had a matching transmitter. Plus the 75S-3 was known to have excellent sensitivity on all bands it covered. The Sherwood Engineering site rates it at better than -140 dBm IIRC.
G3RZP: "I live in an area where as far as I can tell (because the measurements are of necessity rather crude) the noise is a few dB below the ITU R recommendation P.372-8 level for a quiet rural area."
That's great! Unfortunately, in my 42 years as a ham I've always lived in rather noisy suburban locations, where the classic test of "how much does the noise drop when you switch to the dummy load" is easily passed.
G3RZP: "Most people run into trouble with the 6EH7 because they run it at full gain, so the mixer overloads A low plate load resistance and a 27ohm unbypassed cathode resistor help linearity without sacrificing too much noise figure – which you can afford to do at HF, as you say."
In the Southgate Type 7, the 6EH7 RF stage is stabilized by extensive bypassing, shielding, low plate load resistance and not being asked to run at full gain. It has its own independent gain control that is set for optimum performance as needed.
The Type 7 also uses a 6EH7 as the first IF stage because most other remote-cutoff tubes like the 6BA6 were too noisy.
G3RZP: "have you looked at using two tuned circuits in an image rejection arrangement?"
Yes, but I went with a more-conventional approach. With an RF stage I have two tuned circuits anyway.
G3RZP: "an E88CC/6DJ8 in the HRO rebuild I have planned. An absolute basket case that I bought cheaply and not even mechanically complete, the intention is to try a 6EH7 RF amp, with two tuned circuits in image reject arrangement ahead of it and a 6DJ8 mixer with a 6C4 oscillator, followed by 1.4 MHz crystal filters ( of which I just happen to have a large selection in the junk box)."
With the multiple-gang capacitor of the HRO, why not just have multiple tuned circuits in the front end? Say, a single tuned circuit to match the antenna to the grid of the RF stage, then a double-tuned circuit between RF stage and mixer. That way, the mixer only has to deal with a relatively small number of kHz at its grid.
I may have the same filters you do, or something similar. Back in the 1980s I acquired some 1.4 MHz 500 Hz wide filters at a hamfest for fifty cents each. They were cheap because the seller didn't have specs and didn't know if they were any good. A friend found the specs and they were 8 pole filters. I use two of them in the Type 7, one between mixer and first IF stage, and one between the first and second IF stages. They were made by Alpha Components and are rectangular grey boxes about 3 inches long and 1 inch square cross-section.
73 de Jim, N2EY
|
| |
|
RE: 7360 noise
|
Reply
|
|
by G3RZP on April 16, 2009
|
Mail this to a friend!
|
Hi Jim,
Looking at the 7360 circuit etc, it’s certainly fair to apply the noise equations for a tetrode or pentode to the electron stream after the screen grid and before the deflection plates. Now if the 1500 ohms from that really applied, we have to answer the question as to why the SS1R was so insensitive, as it should have done much better than 10dB SNR for 0.5 microvolt if the enr was 1500 ohms. (Let’s leave aside the fact that such sensitivity isn’t actually useable at HF)
The same applies to the 6JH8 receiver, too. I have a feeling that the only answer is going to be actually do some measurements. Possibly the SS1R was a bit too expensive because of the mechanical complexity.
The Marconi Marine Pennant channelised HF receiver for maritime SSB used a 6EH7 RF stage and a 4 diode ring bridge mixer. With a 27 ohm unbypassed resistor in the cathode, they still had a noise figure of about 8dB – again, far better than really needed. At the same time, it far exceeded the IMD requirements of the time, which were -4dBm intercept point.
The idea for the HRO is two coupled tuned circuits in the image reject configuration (which gives up to 30dB more image rejection than just two tuned circuits), followed by a 6EH7 amplifier, a tuned circuit, a 6DJ8 mixer, 6C4 oscillator and a after the filters, a 6EH7 amplifier – low noise being required. The NF should still be adequate with two circuits prior to the 6EH7. I have 1.4MHz crystal filters of various types – for this project, 8 pole USB and LSB, 6 pole 7kHz for AM, 6 pole 1.2KHZ, 600Hz and 100Hz for CW, so 6 in all.
My article in NCJ last year showed that measurements indicate that about 100dB instantaneous dynamic range is all that's needed, even on 7MHz, so that in fact, phase noise is potentially more of a problem for a reasonable design than is IMD. So I'll have an antenna attenuator, probably switchable in 5 dB steps up to 30dB
I looked at the Pullen mixer. There isn’t a noise analysis published and the statement about the linearity of the ratios of the gms isn’t, at least to me, intuitive. I can see that the negative feedback inherent in the first stage acting as a cathode follower should enable a fair amount of signal handling, and grid leak biasing the second stage will tend to make performance rather less dependent on oscillator signal level. But it does look as if the second stage is really only a conventional triodes mixer in which conversion transconductance is 0.25gm, and enr is 2.5/0.25gm. Even so, that doesn’t mean it can’t provide adequate performance of course. It isn’t that different to the one in the 1964 Single Sideband Fundamentals and Circuits, where the 12AU7 needs 9 microvolts for 10dB SNR while giving a 40dB IM ratio at 2 volts. Comparison is very difficult when you don’t know the step ups involved and the noise contributions from them – plus the frequency because the induced grid noise comes into play. Interestingly, noise in CMOS RF amplifiers has an induced gate noise component at UHF, caused by exactly the same mechanism – transit time!
73
Peter G3RZP
|
| |
|
RE: 7360 noise
|
Reply
|
|
by N2EY on April 17, 2009
|
Mail this to a friend!
|
Hello Peter,
An article you've probably encountered was in QST about 1971, called "An Experimental Receiver for 75 Meter DX Work" by W1KLK (SK, unfortunately).
This RX used some pretty interesting brute-force methods. The front end had four tuned circuits at the signal frequency - two between the antenna and RF amp input, two between the RF amp output and the mixer.
The RF amp was a 7044 dual-triode operated in grounded grid. It ran at low gain; its main purpose was to overcome the losses in the four tuned circuits. The mixer was a 7360 operated single-ended, followed by a mechanical filter probably from an R-390A. No numbers presented but an interesting idea.
You might want to take a look at:
http://amfone.net/Amforum/index.php?topic=16720.0
W1VD tested an SS-1R last year and reports the following numbers on 40 meters: (On 40, the SS-1R is single conversion):
Stock (6V p-p) LO:
MDS: -136 dBm
Blocking (20 kHz): 120 dB
Two tone (20 kHz): 89 dB
HP8640B external LO (10V p-p)
MDS: -138 dBm
Blocking (20 kHz): 129 dB
Two tone (20 kHz): 94 dB
HP8640B external LO (14V p-p)
MDS: -138 dBm
Blocking (20 kHz): 132 dB
Two tone (20 kHz): 94 dB
Pretty interesting numbers for a 45+ year old receiver! W1VD says the two-tone IMD may be limited by the IF filters used in the SS-1R.
Your HRO rebuild sounds very interesting. I didn't know that the image rejector circuit would give so much more rejection! Would it be better than three tuned circuits?
One possible way to avoid having a "fussy" input circuit would be to put the image rejector between the RF amp and mixer, rather than at the antenna. That way it doesn't have to deal with changing antenna impedances. Just an idea.
On the Pullen mixer, it's not clear to me that anyone has done a really solid analysis of how well it lives up to its claims. Particularly whether it is quieter/higher conversion gain/better dynamic range than, say, a 6JH8 in balanced configuration.
Where I think the Pullen mixer can really shine is in souping up older receivers that have very noisy pentagrid mixers, such as the WW2 surplus BC-342N. Won't make it a world-beater by any stretch of the imagination but it would be a big improvement over the stock 6L7 mixer, with a no-holes 100% reversible mod.
What would really be interesting is for someone with the equipment and know-how to build a very simple one-band testbed hollowstate receiver, (something like the G4OEP design) and try various circuits and components to get real numbers for various combinations by measurement using standard modern methods.
---
An idea I've been tossing around the past few years is to build a successor to my current rig, using the same 1.4 MHz filters but a different front-end scheme. Something like this:
The design would only cover 80, 40 and 20 meters, CW, because those are the bands of interest here. The tunable LO would cover 5.1 to 5.4 MHz, and for transmitting would be mixed with the signal from either an 8.9 MHz (80/20 meters) or 12.4 MHz (40 meters) xtal oscillator to cover the three bands. (Since this is a CW-only rig, full coverage of 80 and 20 is not needed).
The receiver would start out with a double-tuned input circuit, then a high-performance mixer that uses the same tunable LO as the transmitter section. Output would be on either 8.9 or 12.4 MHz, depending on band, and a high-Q double-tuned IFT for either frequency would be switched in as needed.
The 8.9 or 12.4 MHz first IF would then be converted to 1.4 MHz by means of a second high-performance mixer, with injection from a crystal oscillator at 10.3 or 13.8 MHz.
I know it sounds like a convoluted design, but it's based on parts I have on hand, simplicity and the desire to build a transceiver rather than a receiver.
The obvious question is, what to use for mixers? If the 7360 or 6JH8 can do as well as W1VD's measurements indicate, it's an obvious choice, but would it really be better than, say, a 6ES8 or 6DJ8 in Pullen configuration?
73 de Jim, N2EY
Thanks for an interesting discussion and some ideas.
|
| |
|
RE: 7360 noise
|
Reply
|
|
by G3RZP on April 18, 2009
|
Mail this to a friend!
|
Jim,
Sorry for the delay in replying – I’m travelling. Right now in Phoenix, tomorrow to San Diego, next Saturday to Washington DC, and the back via the south of France for another meeting
The Pullen mixer might be a bit low gain for replacing a 6L7 or similar.
The image rejection coupling of tuned circuits isn’t widely known, and only works over a narrow band, but certainly offer some improvement for only an extra winding on the coil in the second tuned circuit. So three tuned circuits, two of them coupled with image rejection, would certainly help above 15MHz
The SS1R figures show the limitation of the LO injection method used. That’s the advantage of the G3PDM PLL approach, although a low noise mixer in synthesizer path might be helpful. Not sure what loop bandwidth he ended up with.. Of course, at the end of the day, the practical limitation will be the cleanliness or otherwise of the tx causing the problem! As I said earlier though, even in Europe, the requirement is for about 100dB of dynamic range – more doesn’t gain you anything.
The transceiver sounds interesting, but a PLL injection loop and going straight to 1.4 would avoid the requirement for too good mixers in the rx path.
Good luck with it, anyway.
73 Peter G3RZP
|
| |
|
RE: 7360 noise
|
Reply
|
|
by N2EY on April 20, 2009
|
Mail this to a friend!
|
Peter,
So near yet so far - I'm only a few hours north of Washington DC!
I think that a properly-chosen Pullen mixer would have at least as much gain as a 6L7. My tube books list the 6L7 as having 350 umhos of conversion gain, which shouldn't be too hard to beat with, say, a 6SL7.
Yes, if the receiver dynamic range is high enough, the cleanliness of the big signal becomes the main issue. More than once, I've encountered situations on Field Day where what was originally thought to be overload from a same-band rig turned out to be transmitter noise.
G3RZP: "The transceiver sounds interesting, but a PLL injection loop and going straight to 1.4 would avoid the requirement for too good mixers in the rx path."
The big problem there is that the use of 1.4 MHz in a transmitter puts the fifth harmonic on 7 MHz and tenth harmonic on 14 MHz. There's also the complexity of an all-tube PLL compared to the two-receiving-mixers approach.
What I'm really trying to do with it is build a decent rig with only parts on hand.
TNX and have a good trip.
73 de Jim, N2EY
|
| |
|
RE: 7360 noise
|
Reply
|
|
by K3IMW on May 15, 2009
|
Mail this to a friend!
|
I've been tinkering with a variant of G4OEP's elegant and high-performance design for a while. I can't comment on the noise figure of the 6JH8, but I have measured the conversion transconductance. The rule of thumb that gc is 1/3 of gm applies reasonably well here, provided that enough drive is applied to the deflector electrodes to switch, way, 90% of the beam current from plate to plate.
I think that it is correct from an ENR viewpoint to treat these beam deflection tubes as pentodes with a steered beam - in other words, there is no partitioning noise term due to the deflection electrodes.
At present I am using the Si570 as LO, with a video amplifier opamp to provide wideband drive to the deflectors with no tuned circuits. This doesn't provide enough voltage swing to obtain maximum gain, but it does make for a very simple multiband front end. This also means I can't use Andy's neat idea
of varying the LO drive as part of the AGC chain (this allows him to get adequate AGC control with only a single IF amp stage). I plan to try the G3PDM phase-locked oscillator approach with modern components to allow a totally hollow-state implementation.
The 7360's characteristics are less linear than the 6JH8 but it requires less drive (I have a sneaking suspicion that they are the same tube, operated in different voltage regions).
73, ian K3IMW
|
| |
|
RE: 7360 noise
|
Reply
|
|
by G8HQP on May 30, 2009
|
Mail this to a friend!
|
Hi,
I stumbled upon this thread a few weeks ago. I've been doing some thinking about valve noise for some months, and the thread stimulated further thinking.
To cut a long story short, I think the enr of a 7360 will be a minimum of 3k and could be as high as 14k. The 3k figure comes from the pentode calculated enr of 1500, then doubled because the output receives noise from both the signal and image noise bandwidth. Unlike a normal pentode mixer, there is no shot noise or g2 partition noise contribution directly at the IF provided the output circuit is balanced, as will usually be the case for a beam-deflection mixer. This also means that the 3k enr will not increase at lower LO levels, as the noise and signal vary together.
The higher figure of 14k would arise if there is a partition noise contribution from each electron having to randomly choose which anode to arrive at. I suspect that this choice only has to be made by some electrons, in which case the enr will be somewhere between 3k and 14k. This additional 11k enr will depend strongly on the LO level, disappearing altogether for full-level square-wave LO but getting worse for low level sine-wave LO.
See http://myweb.tiscali.co.uk/g8hqp/radio/vmixernoise.html for further details.
As has been said, the final receiver sensitivity depends on the input arrangements from the antenna to the grid.
regards
Dave G8HQP
|
| |
|
RE: 7360 noise
|
Reply
|
|
by N2EY on June 1, 2009
|
Mail this to a friend!
|
G8HQP writes:
"The 3k figure comes from the pentode calculated enr of 1500, then doubled because the output receives noise from both the signal and image noise bandwidth."
I don't think that's true.
The noise figure of an amplifier or mixer is simply the noise it generates itself, not noise already present at the input.
Where noise at the image frequency becomes a concern is when the receiver input selectivity is such that a significant amount of *external* noise at the image frequency gets to the mixer. (I have read that this is the case in some radio-astronomy and radar applications where the selectivity before the mixer is low). But what happens in that case is degradation of the *system* noise performance, not the mixer noise figure.
In HF receivers, the image and IF rejection should be such that those things aren't a problem.
IMHO the only realistic way to get a handle on the noise figure of a beam-deflection mixer is to build a receiver using one and measure the performance.
73 de Jim, N2EY
|
| |
|
RE: 7360 noise
|
Reply
|
|
by G8HQP on June 2, 2009
|
Mail this to a friend!
|
N2EY writes: "The noise figure of an amplifier or mixer is simply the noise it generates itself, not noise already present at the input."
I agree. The image noise I am talking about is not external image noise, coming in at the grid, but image noise generated internally. Think of a 7360 as a pentode amplifier, closely coupled to a mixer; in fact, so closely coupled that there is no opportunity to interpose a filter. The electron stream in the 'pentode' will consist of DC, the amplifed input signal and internally generated noise. The mixer will then multiply this signal by the LO. The result is the usual sum and difference frequencies. As it is a balanced mixer there will be no output components at the original signal frequency. However, it is not an image-cancelling mixer so the output receives two contributions: from the wanted signal plus accompanying noise, and the image. If there is good image filtering before the 7360 then the image consists solely of internally generated noise which will be at the same level as the internally generated noise around the signal frequency. As these two noise bands are at different frequencies they will presumably be uncorrelated so will add like power rather than voltage i.e. 3dB extra. Hence the enr is doubled.
Exactly the same effect happens in a conventional unbalanced triode/pentode mixer, except there you also get a noise contribution directly at the IF. This is why the numerator in the formula for enr changes from 2.5 to 4 when going from an amp to a mixer.
It took me a while to work this out, so I'm not surprised it seems surprising!
If we knew the voltage step-up, unloaded Q etc. of existing receivers for which reliable measurements have already been taken it ought to be possible to estimate the enr of the 7360. If people get better results with higher LO levels then that would confirm that the 'mixer' part is introducing its own partition noise. Some figures already seem to show this.
73s
Dave G8HQP
|
| |
|
Email Subscription
You are not subscribed to this topic.
Subscribe!
My Subscriptions
Subscriptions Help
Check our help page for help using
Forum, or send questions, comments, or suggestions to the
Forum Manager.
|
|
|
