Re: Final final 1kHz oscillator

On 23/11/2024 3:32 am, Edward Rawde wrote:
 > "Bill Sloman" <bill.sloman@ieee.org> wrote in message news:vhp713$12bnt$2@dont-email.me...
 >> On 21/11/2024 1:00 am, Bill Sloman wrote:
 >>> On 20/11/2024 2:03 pm, Edward Rawde wrote:
 >>>> "Bill Sloman" <bill.sloman@ieee.org> wrote in message news:vhjj2v$24eu4$3@dont-email.me...
 >>>>> On 20/11/2024 1:29 pm, Bill Sloman wrote:
 >>>>>> On 20/11/2024 12:59 pm, Edward Rawde wrote:
 >>>>>>> "Bill Sloman" <bill.sloman@ieee.org> wrote in message news:vhibce$1t7v2$1@dont-email.me...
 >>>>>>>> On 18/11/2024 2:58 pm, Edward Rawde wrote:
 >>>>>>>>> "JM" <sunaecoNoChoppedPork@gmail.com> wrote in message news:n7iijjdeqecl0kmub0bq5in0dbm60m7qam@4ax.com...
 >>>>>>>>>> On Thu, 14 Nov 2024 11:14:28 -0500, "Edward Rawde"
 >>>>>>>>>> <invalid@invalid.invalid> wrote:
 >>>>>>>>>>
 >>>>>>>>>>> "JM" <sunaecoNoChoppedPork@gmail.com> wrote in message news:t5fajjdteskfftvkf84iqsp2vc4b9ta5kj@4ax.com...
 >>>>>>>>>>>> On Fri, 8 Nov 2024 15:43:41 -0500, "Edward Rawde"
 >>>>>>>>>>>> <invalid@invalid.invalid> wrote:
 >>>
 >>> <snip>
 >>>
 >>>>> I've no idea why you are using the LT1994. The circuit doesn't have a common mode problem, so why are you using an op amp
 >>>>> designed
 >>>>> to deal with one?
 >>>>>
 >>>>> This is sci.electronics.design not sci.electronics.incremental-development.
 >>>>>
 >>>>> And the six diode "stabiliser string" is nuts. If you need a 3.6V reference voltage there are lots of ways to do it with more
 >>>>> precision and better temperature stability.
 >>>>
 >>>> In that case please show a circuit with better performance.
 >>>> It may be that some of the circuits JM posted can do better but if so then why are you using a circuit with so many components?
 >>>> My circuit has far fewer components than yours.
 >>>
 >>> But you don't seem to be able to tell us what they do.
 >>>
 >>> I think I've found my conceptual problem with my circuit. Tweaking the gain around the three-amplifier ring tweaks both amplitude
 >>> and frequency - with more gain a lower frequency signal can still propagate around the ring.
 >>>
 >>> I've got to find a mechanism that will separate amplitude from frequency. My copy of Williams and Taylor on electronic filter
 >>> design may get perused again.
 >>
 >> I found a simpler solution - copy the relevant arrangement in John May's circuit. It did work - after a fashion - but as I got
 >> closer to getting it to a state where it could do what I wanted, the circuit got less and less willing to simulate.
 >>
 >> I suspect an accumulation of typo's in component values - I do try to go the through the schematic to find and purge them. But the
 >> last few passes haven't shown up anything. Frustrating. My father's advice in similar situautions was to "drop it in drawer for
 >> six months, then take another look". It has worked in the past.
I found that adding a couple of 14uH ferrite beads around the transistors and the FET stopped the simulation dropping out after getting stuck on a too-short time step.
The current version isn't simulating all that fast - I let it run over-night and the amplitude control feed-back loop turned out to have been underdamped to the point of instability - it kept on hitting the rails, recovering and overshooting back into them. The current version - with more damping - is now on it's second millisecond.
 > I also found a simpler solution. Taking on board advice from JM and others.
 >
 > The circuit below does 0dB into 600 ohms and it only takes about 20 minutes to complete the simulation on my computer.
 >
 > When it's done, select a sample of about 1 second near the end and FFT.
 > Select Use current zoom extent and Blackman-Harris window.
 >
 > It will say all harmonics are more than 120dB down.
 >
 > I'm not saying this level of performance is achievable or measurable in reality so I don't see any point simulating further.
 >
 > The actual distortion in reality will likely be that of the op amps so choose the lowest distortion op amp you can find.
 >
 > Oh and if you need to know the exact function of any of the 21 components in this circuit just ask.
 >
 > I'd feel embarrassed to have produced a circuit using over 70 components which only claims 65dB down on harmonics.
Of course you would. The point of producing the circuit is to find out what it can do, and change it to make it work better. That way you get to understand what the circuit is doing and why it is doing it, which isn't your strong point.
 > Does anyone know how to change the default trace in LTSpice?
There isn't one in the version I downloaded. When I start a simulation I get offered a blank display, and have to select a trace to be displayed before I can see anything.
 > When I click Run/pause it shows the wrong node so then I have to Delete This Trace and click output.
 > How do I make it default to output?
Beats me.
--
Bill Sloman, Sydney