This is all very weird.
When I got it right, nothing happens with the scope connected to the 47k but without you get the huge bump?
That would imply that the probes capacity is enough to stop the bump.
What happens just for testing when you place instead of the probe a similar cap value of some 25pF over the 47k.
What about your Cart, many MM Carts have a connection from one pin to their body, thereby electrically connecting it via the arm to the TT frame.
In that case your Cart is not floating.
Would it be possible to show in a sketched diagram exactly how the connections are via your interconnect cable to TT frame and Cart and whether your TT is connected to PE through its main cable.
And is the power supply of your pre amp also connected to PE ?
Hans
Tell it to todays audio gurus. There are more modern popular designs presented at diyaudio.com without RF filters than with RF filters it seems. It would be interesting to check how many threads less there would be when standard practices of decades ago would be reintroduced. The stupidity to present products that don't have any problem amplifying 10 MHz....
Use 40 to 110 MHz capable parts, use no filtering, have Bluetooth/LED lighting/SMPS/wireless internet/DECT/smart phone in abundance (or the neighbours), get MHz in the chain. The cell phone test at a few centimeters distance is a nice test. Make sure to lower volume to very low level
The OPA1656 is not a speaker-eater. I think it is absolutely not fair to blame an excellent product because of the imperfections of a certain design it is used in. If it and other opamps malfunction in the design then the design is to check. Decoupling is best done with SMD caps very close to the chip so a real PCB. In my experience local low noise regs also work better than the old distributed power design.
Last edited:
That's exactly right Hans. Whenever there is a Blue trace it's the TT input where the scope probe tip and its Gnd lead are attached to RCA center and RCA shield, resp. Yellow is always RIAA output. Yellow only is with TT signal straight in - just like the first time I tried my box.
> The TT has a 2-prong AC cord so no PE.
> There is no connection between the Right channel shield and the TT Gnd wire.
> There is roughly 9ohms between the Left channel shield and the TT Gnd wire. Hmmm...
> My scope manual says - Input impedance = 1Mohm // 20pf which no doubt is internal at the BNC connector.
> When I measure just the disconnected scope lead I get ~35pF.
I used an empty RCA male connector and installed a 15pF polystyrene cap inside and plugged that into one female jack on the 1-male to 2xfemale adaptor. Then plugged the Right channel TT cable into the 2nd female jack. Then the single male into the box. So now TT signal enters the RCA adaptor, sees 15pF from center to shield and then the 47K a few inches inside the box on the PCB. That's it.
What this does is to transform the whole situation. Gone are the multi-volt "speaker-eater" convulsions and back are the "with scope probe" many tens of mV "docile" OPA627 results for Lift and Lower - the latter always being the greater.
Attached are today's scope shots labelled the same way as yesterday's + two shots of my setup. Unbelievable. I believe I can live with this solution.
Attachments
So the 4562/49702 seems to be the outlier. Could there be reason connected to the fact that the data sheets (identical) indicate that Ib and Ios are uncompensated? The lack of a "+-" for both those values usually means non-compensated. Have also read that sometimes the input bias type could be compensated but not indicated. Just throwing that out there.
Tell it to todays audio gurus. There are more modern popular designs presented at diyaudio.com without RF filters than with RF filters it seems. It would be interesting to check how many threads less there would be when standard practices of decades ago would be reintroduced. The stupidity to present products that don't have any problem amplifying 10 MHz....
Use 40 to 110 MHz capable parts, use no filtering, have Bluetooth/LED lighting/SMPS/wireless internet/DECT/smart phone in abundance (or the neighbours), get MHz in the chain. The cell phone test at a few centimeters distance is a nice test. Make sure to lower volume to very low level
The OPA1656 is not a speaker-eater. I think it is absolutely not fair to blame an excellent product because of the imperfections of a certain design it is used in. If it and other opamps malfunction in the design then the design is to check. Decoupling is best done with SMD caps very close to the chip so a real PCB. In my experience local low noise regs also work better than the old distributed power design.
I hear you loud and clear re my house's infestations with all manner of RF. When at the bench and scoping some small signal I many times say - what the heck is THAT? As in, some hi-freq burst flying across the screen, etc.
I apologize for the "speaker-eater" comment. But, at the time I had built this RIAA design found in a TI data sheet for a hot-rod TI audio op amp and it did almost eat my subwoofers. I was running two, 80W Class A tube amp monos at the time for the mains. God knows what would be left of my mains had I been running the SS 300W/chan Bryston 4B ST2...?
Any way there appears to be a solution I can live with so can't wait to tidy-up and listen to this wonderful OPA1656.
If I were you, I'd like to know how close it is to the edge of the cliff with 15 pF, especially since a capacitor without resistor theoretically detunes rather than damps the cable resonance. Does it go wrong again with 12 pF? Does a resistor in series with the capacitor (such that the RC time constant is around 3 ns) improve matters?
JRA, I had it once when trying the Power LAN devices I got for testing and then heard strange noises on my speakers. RF is an issue these days and sometimes getting it solved it quite hard. The ham guys have the knowledge as RF is not that familiar to most involved in audio. One notorious one was a 6V SMPS that was delivered with a device that stopped a media player from playing. Also speaker cables picking up garbage is common.
Buildings where the landlord has the telecom companies having smart phone transmitters installed on the roof are among the most difficult. I think one better changes hobby then
Glad that the cap solved it.
Buildings where the landlord has the telecom companies having smart phone transmitters installed on the roof are among the most difficult. I think one better changes hobby then
Glad that the cap solved it.
Last edited:
A polyethylene filled cable of 1 metre has a quarter-wave resonance around 50 MHz. Your cable is a bit shorter, but there may be some internal wiring between the shorting switch and the cable, so 50 MHz is still a reasonable guess.
For a given capacitance, an RC series network will give you the largest conductance (smallest equivalent parallel resistance, so best damping) when the value of the series resistor equals the reactance of the capacitor at the frequency of interest. (That's something I calculated for work decades ago and remembered, as I use it frequently.) That is, R = 1/(2 pi f C) or RC = 1/(2 pi f). Fill in 50 MHz and you get RC ~= 3.183 ns. Since it's all based on guesses anyway, I rounded that to 3 ns.
This is all for an RC series network shunting the input to damp the resonance. I expect you can use larger resistors when you go for an RC low-pass filter, but I haven't written out the equations for that. My gut feeling is that you can use any value from 3 ns/C up to the largest value you find acceptable for noise when you make an RC low-pass.
For a given capacitance, an RC series network will give you the largest conductance (smallest equivalent parallel resistance, so best damping) when the value of the series resistor equals the reactance of the capacitor at the frequency of interest. (That's something I calculated for work decades ago and remembered, as I use it frequently.) That is, R = 1/(2 pi f C) or RC = 1/(2 pi f). Fill in 50 MHz and you get RC ~= 3.183 ns. Since it's all based on guesses anyway, I rounded that to 3 ns.
This is all for an RC series network shunting the input to damp the resonance. I expect you can use larger resistors when you go for an RC low-pass filter, but I haven't written out the equations for that. My gut feeling is that you can use any value from 3 ns/C up to the largest value you find acceptable for noise when you make an RC low-pass.
With all due respect I shy away from upside-down TT disassembly, finding metal to connect a wall PE to, etc. Aka PITA.
Is there another way to try to inhibit this knee-jerk reaction the 1656 has to otherwise tiny input insults or "everyday" life? Some combo of stimuli are setting-off the 1656 like it's highly explosive - in a circuit provided by its Maker.
As I suggested before, it seems to want to fix an input aberration that some other op amps shrug-off.
PE is a means to shunt RF to. It is also quite safe to use PE. A thin wire with some tape to a metal part for testing is not too hard I guess. Does the amplifier have a 3 prong plug connected to a PE socket?
(just looking for the simple things besides RC filtering). Its maker probably had it developed in a lab
(just looking for the simple things besides RC filtering). Its maker probably had it developed in a lab
Last edited:
Above 50MHz or so a 1.5m PE connection will probably shunt not very much. At this frequency that equals one quarter wavelength, a perfect antenna. At least a widely used one
Monopole antenna - Wikipedia
Monopole antenna - Wikipedia
No, it will be beneficial for the lower frequencies, but isn't that connected by the ground wire anyways? Sorry, no turntable guy here, my chains are all digital.
What I wanted to say - for RF like cell phones, WiFi etc you need to take other precautions like rc filters at the inputs for example
What I wanted to say - for RF like cell phones, WiFi etc you need to take other precautions like rc filters at the inputs for example
Why is it thought that the striker for this explosive behavior is RF in the air?
The trigger is a slide switch which shorts together or un-shorts the two ends of an unexcited coil, at rest, doing nothing. That is all it takes for the 1656 to decide it's time to blast off to the moon - albeit in these particular surroundings.
The trigger is a slide switch which shorts together or un-shorts the two ends of an unexcited coil, at rest, doing nothing. That is all it takes for the 1656 to decide it's time to blast off to the moon - albeit in these particular surroundings.
PE is a means to shunt RF to. It is also quite safe to use PE. A thin wire with some tape to a metal part for testing is not too hard I guess. Does the amplifier have a 3 prong plug connected to a PE socket?
(just looking for the simple things besides RC filtering). Its maker probably had it developed in a lab
2-prong.