Tweet
Conduction angles:
-
-
Transient responses during a slope:Comment
-
R1 must be tuned properly to maintain amplitude stability and signal purity. It should be made as 22 ohm in series with a trimmer potentiometer of 50 - 100 ohm. For the purpose of tuning, I indicated a TP on a schematic. Oddly enough, the same procedure sould be applicable to all other single-ended oscillators with transistor in output. So when in doubt, try this method.
Perfect value is obtained when the bump that forms in a middle of a conduction period does not cross 0V. It may get a bit flat, but the transistor current should not go reverse.Comment
-
Hey, lighten up guy, I was being facetious, OK? You do know the meaning of the word?Originally posted by rickb View Post
As for quelling Davor's interest in returning to the forum, oh please, give me a break.Comment
-
Facetious or faecal, I don't care. After this many years on different fora one grows a thick hide.
But the truth is I'm much more busy lately than ever before, and if you suddenly miss my posts, it's because I'm not here.
Anyway, electronics is my mental hygiene, so I'll drop by occasionally when I make time.
The next topic will be an in-amp front end capable of adopting to any kind of coil configuration, including shieldless balanced ones. So as long as the coil is built for around 13 kHz operation, you'll be able to hook it to this front-end.
After that comes a double balanced IQ mixer, and phasing circuit to run it, with the idea is to obtain a fully balanced IQ quadruplet for running VDIs etc.Comment
-
Bring it on!Comment
-
Don't hold your breath, it will be a process. My idea is to make a platform that works on its own, with basic functionality of IGSL: 2 overlap-able proportional tones - non-Fe, Fe, and foil in between. But tuning will be simplified, and there will be TP-s where needed, and positions to hook a VDI with fixed 0/90? phases, and a position to add a pinpoint function with or without VCO, etc. And no convoluted circuitry whatsoever.Comment
-
What are your ideas on i/q demod ?Comment
-
You may or may not recall, I was one of the first persons to express my gratitude, for you returning to the forum after a long absence.Originally posted by Davor View Post
P.S I believe the expression is "thick skinned", as only large animals have hides.Comment
-
Precisely. Forum animals grow large, and have thick hide. Or go extinct.Comment
-
I intend to use CD4052 as a double balanced IQ mixer. It will produce 0, 90, 180, and 270 degrees products, so hooking up VDI will be easy. Phasing network will be slightly different than the most classic detectors'. Feeding 4052 requires 45 degree advance, as its conduction angle is 90 degrees, and half of that is 45. Classic detectors have mixers with 180 degree conduction angle, so phase advance is 90 degrees (half of 180). More about that will come soon.Originally posted by rickb View Post
A big advantage of IGSL is that its 3 distinct sound qualities suffice for 99% of target identification. You may get the remaining 1% by means of a VDI. But in terms of speed and ease of use, there are handful of modern detectors that can do it this fast and straightforward. In my use I found that the knobs for adjusting Fe and non-Fe discrimination on the front panel are there mostly to annoy me, because I often turn them by accident, while adjusting GB, and have to tune them back to their proper and constant positions. In other words they may as well be fixed. I'll leave both options: front panel potentiometers, and on board trimmers you set once and forget.
So initial tuning with a new coil will be:
1. setting a 0/90 degree constellation by adjusting only a 45 degree trimmer, using a ferrite pouch,
2. setting discrimination trimmers to obtain a third overlapped tone for foil, using a lump of foil,
3. setting a GB for local terrain by adjusting a GB potentiometer, bobbing a coil above ground.
Further tunings:
- setting a GB for local terrain by adjusting a GB potentiometer, bobbing a coil above ground.
(nothing else)Comment
-
Back to the drawing board. The Rx preamp will be a fully differential amplifier, in aperiodic configuration. A few details, and a stream of thought...
Instrumentation amplifier has a fully differential input, but a single ended output. As I need a differential output for a double balanced mixer, I'll simply omit one stage and get a fully differential amplifier. Because it may work equally well for any kind of input topology, I'll be able to use it with just about any kind of coil there is, including centre-tapped shieldless ones that I prefer.
Aperiodic means non-resonant. By introducing a resonance, as most classic detectors have, it is possible to improve input referred noise by ~3 dB, shifting a resonance away from the working frequency without getting into trouble with phase shift. It is possible to improve noise by some more, but only by going to the very resonance. But there are two problems: noise is limited by the coil resistance, and the phase shift is all over the place. So scratch that.
Because nowadays opamps came close to the noise limit of the very coil, why not discarding the resonance concept altogether?
Advantage of aperiodic front end is that every coil you hook up will work immediately, without any tuning. Plug and play.
Disadvantage of a fully differential input is 3 dB more noise than a single ended input. I think it is worth it.
Because turning an aperiodic front end into a resonant one is only a single capacitor away, I'll leave the choice to the individual builder. If you have more than one coil, you are better off leaving it aperiodic. If you have only one coil, and want to squeeze the most out of it, then play with capacitors. Your choice.Comment
-
Moving on.
After some consideration, I opted for a middle path, basically an aperiodic preamp, but with heavily damped frequency-dependent parts.
The prefered IC here is LME49720. In PDIP form it has a same pinout as NE5532, and many others. But LME49720 has a far superior distortion.
It is fully differential in a sense that it will happily work with differential or single ended coil, producing differential output with only 2% error (gain is ~50) in case of single ended coil. Not bad at all.
A few details...
Gain is about 50 per output, and hence differential gain is about 100.
R1 is the main damping resistor. It defines the impedance the coil sees. It has basically the same purpose as the damping resistor in PI machines. C1R2 are a "damped resonant capacitor", which is not so much resonant as it is good for fixing the higher frequencies. R3-R6, and C2-3 take care of biasing and act as a high pass filter to reduce mains hum entering the mixer.
C4 is to reduce RF influence.
R7 not critical, should be metal film, but hey, aren't they all, and C5 should be bipolar.
R8 and R9 should be from E24 or better, hence odd values, and preferably the same within 1% tolerance. It is OK to put 1k5, or 1k8 from E12, but check their resistance first.
C8 and C9, with R10 and R11, and mixer input impedance form a high pass to further reduce hum, but also to eliminate any offset by preamp.
I'll send LTspice files in the next mail.Comment
-
C5=10uF non polarized as "bi-polar"?
Comment
-
Yup.
It is quite possible to arrange a separate bias for keeping electrolytic capacitors properly polarised, but the non-polarised or bi-polar elcos are not that hard to come by, and provide far less distortion than polarised elcos, and much less than ceramic disc ones.
Are you interested in a circuit that would take care of elco polarisation? It has to be low noise, and low impedance, but according to the late Cyril Bateman's measurements, you are far better off with bi-polars than back to back polarised elcos, and they are not that difficult to come by. See: http://waynekirkwood.com/images/pdf/...ap_Sound_5.pdf
As Cyril Bateman died in 2015, and his works are truly remarkable, some people keep those in collections for safekeeping. Say: https://linearaudio.nl/cyril-bateman...sound-articlesComment
Comment