colpits AGC stability problem

The amplitude in the MC1648 is barely a transition width P-P, so "switching" is not the right description. If you can find an old data sheet or ap-note the spectrum is shown.

...Jim Thompson

I don't understand your terminology, "barely a transition width P-P". That makes no sense.

I have used the 1648 and have the datasheet. The schematic is shown in Fig. 5. It is a Butler configuration.I am quite familiar with the Butler and it's limitations. See my patent #4533881 at

The spectrum for the 1648 was ok back when the device was designed, but those days are long gone. Viewing from today's technology, the 1648 is perhaps the all-time worst oscillator design ever to go into production. For example:

1. As a Butler, the 1648 switches current to the tank at the zero crossings. In other words at a 50% duty cycle. This is the worst possible point for phase noise, as discussed in Lee and Hajimiri.
2. The direct connection from collector to base of the tank transistor severely limits the maximum tank voltage achievable. This also degrades the noise performance.
3. The base-collector junction becomes forward-biased during the negative portion of the tank voltage. This drains energy from the tank and also degrades the phase noise.

I made a simple LTspice analysis that shows these problems. It will be in a reply to this post so those who are not interested do not have to waste time downloading the files.

Regards,

Mike Monett

Antiviral, Antibacterial Silver Solution:

Here's the LTspice file for a 10MHz MC1648 oscillator:

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Version 4 SHEET 1 880 680 WIRE -320 -160 -400 -160 WIRE -256 -160 -320 -160 WIRE -192 -160 -256 -160 WIRE -400 -144 -400 -160 WIRE -320 -112 -320 -160 WIRE 192 -64 64 -64 WIRE -400 -48 -400 -64 WIRE -192 -16 -192 -160 WIRE 0 -16 -192 -16 WIRE -400 32 -512 32 WIRE -320 32 -320 -48 WIRE -320 32 -400 32 WIRE -256 32 -320 32 WIRE 192 32 192 16 WIRE -64 80 -192 80 WIRE 64 80 64 32 WIRE 64 80 -64 80 WIRE -64 96 -64 80 WIRE -512 128 -512 112 WIRE -64 192 -64 176 FLAG 192 32 0 FLAG -512 128 0 FLAG -64 192 0 FLAG -256 -160 Osc FLAG -64 80 Emi SYMBOL voltage 192 -80 R0 WINDOW 123 0 0 Left 0 WINDOW 39 0 0 Left 0 WINDOW 0 42 27 Left 0 WINDOW 3 39 58 Left 0 SYMATTR InstName V1 SYMATTR Value 3v SYMBOL NPN 0 -64 R0 SYMATTR InstName Q2 SYMATTR Value 2N2369 SYMBOL VOLTAGE -512 16 R0 WINDOW 0 39 36 Left 0 WINDOW 3 31 79 Left 0 WINDOW 123 0 0 Left 0 WINDOW 39 0 0 Left 0 SYMATTR InstName V3 SYMATTR Value 1.5V SYMBOL NPN -256 -16 R0 SYMATTR InstName Q1 SYMATTR Value 2N2369 SYMBOL ind -384 -48 R180 WINDOW 0 36 80 Left 0 WINDOW 3 36 40 Left 0 SYMATTR InstName L1 SYMATTR Value 795.7nh SYMBOL cap -336 -112 R0 SYMATTR InstName C1 SYMATTR Value 318.3pf SYMBOL res -80 80 R0 SYMATTR InstName R1 SYMATTR Value 1k SYMBOL res -416 -64 R0 SYMATTR InstName R2 SYMATTR Value 1 TEXT -426 186 Left 0 !.tran 0 250n 0 1n TEXT -96 -176 Left 0 ;'MC1648 10MHz Butler Osc TEXT -432 152 Left 0 !.IC v(Osc) 1.511 TEXT -40 -144 Left 0 ;Tank Q = 50

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Here's the PLT file. Watch the line wrap.

Regards,

Mike Monett

Antiviral, Antibacterial Silver Solution:

It doesn't "switch", it's 'small" signal compared to what you are talking... ~150mV P-P

It doesn't SWITCH already! Do you have any clue about doff pairs?

True enough. I can't do big swings with small supplies and low breakdown devices.

Nonsense, you're clearly not an IC designer. 75mV is hardly "forward-biased".

...Jim Thompson

Except you left out all the important parts. What a dork ;-)

It doesn't operate like you think it does.

READ the patent.

DORK ;-)

...Jim Thompson

-- | James E.Thompson, P.E. | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona Voice:(480)460-2350 | | | E-mail Address at Website Fax:(480)460-2142 | Brass Rat | |

Certainly WOULD be a waste of time... you completely missed the core of this thread ;-)

...Jim Thompson

You are showing why the Butler is the worst possible design for low voltage ic's. The limited voltage swing means poor phase noise. Switching at the zero crossing is the worst place to do it.

As Lee and Hajimiri show, the Colpitts is much better. It is also much better suited to low voltage designs, and can give far higher tank voltage swings than the Butler.

I sure hope you are not still putting 40 year old technology in your current designs.

Regards,

Mike Monett

Antiviral, Antibacterial Silver Solution:

The thread is the Colpitts, not some 40 year old Butler.

Regards,

Mike Monett

Antiviral, Antibacterial Silver Solution:

The title of the thread is "colpits AGC stability problem"

You totally missed the **AGC**

Dork ;-)

...Jim Thompson

Howdja like the instant start? Bet that's the first time you saw that:)

I know the 1648 doesn't run that high. The whole purpose was to show that it can't. There is nothing you can do to improve it. You are stuck with the performance of a 40 year old design, and it is lousy.

Try the Colpitts. It will run at 1 volt or below, where the Butler simply won't work at all.

As my article below discusses, you can also increase the base-emitter capacitor to 5 or 10 times the emitter-ground capacitor.

This has the desirable effect of swamping the junction capacitance even more than usual, which reduces the undesired effect of temperature and bias changes on the oscillator frequency.

It also lets you run at much higher voltage swings across the tank without breaking down the base-emitter junction in the reverse direction. This gives the best possible signal-to-noise ratio in the oscillator, thus minimizing phase noise. It also helps align the collector current pulse with the peak tank voltage, further reducing phase noise.

You can do none of these things with the Butler. There is no place where Rohde, Lee, or Hajimiri would use a Butler in any critical design. Why should you? Regards,

Mike Monett

Antiviral, Antibacterial Silver Solution:

Remember the old saw about having brain fully engaged before putting mouth in high gear?

You slipped a cog ;-)

And you certainly didn't correctly analyze NOR simulate the MC1648 correctly.

Where is it you work?

Or are you currently unemployed ?:-)

...Jim Thompson

Give it up, Mike, you're trying subterfuge because you still don't understand how the MC1648 actually works... or you wouldn't have posted that pathetic LTSpice excuse for a proper analysis.

...Jim Thompson

In message , dated Sat, 26 Aug 2006, Jim Thompson writes

Dofferintial? You must mean 'deferential'.

-- OOO - Own Opinions Only. Try

John Woodgate, J M Woodgate and Associates, Rayleigh, Essex UK

In message , dated Sat, 26 Aug

Jeeves?

-- OOO - Own Opinions Only. Try

John Woodgate, J M Woodgate and Associates, Rayleigh, Essex UK

Yes ;-) I erred and let Mike get under my skin, and was typing too fast... the wife was screaming, "Your dinner is getting cold" ;-)

...Jim Thompson

-- | James E.Thompson, P.E. | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona Voice:(480)460-2350 | | | E-mail Address at Website Fax:(480)460-2142 | Brass Rat | |

Butler? Isn't the resonator connected emitter-to-emitter in a Butler?

...Jim Thompson

-- | James E.Thompson, P.E. | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona Voice:(480)460-2350 | | | E-mail Address at Website Fax:(480)460-2142 | Brass Rat | |

It can be, but doesn't have to. Here's an example:

The Butler is a differential pair with feedback from one collector to the opposing base. The tank can be series resonant, and put between the emitters, or it can be parallel resonant, and go from the base to ground as in the 1648.

I never use subterfuge. I know how the 1648 works. I was shocked to find the signal level on the 1648 is only 150mV p-p. I knew it was bad, but not that bad. With only 150mv, the SNR must be terrible.

Do you know anything about phase noise? It wasn't that important back in 1960, but it sure is today. The specs needed to meet today's communication requirements are very tight, and the 1648 configuration (Butler) simply won't meet them.

For example, here's a 78MHz trimmable crystal oscillator for paging applications. It shows a measured phase noise of -100dBc at 100Hz from the carrier, so the output can be multiplied up to the operating frequency of 470 MHz or 900MHz and still meet the phase noise spec of -75 dBc at 100 Hz offset. Would a Butler do that? No. They use a Colpitts: (111k pdf)

Here's a 5 to 6 GHz VCO for 802.11. They claim it works down to 1 volt. Would a Butler do that? No. They use a Colpitts: (517k pdf)

Precise timing is critical in UWB systems. Here's a 40MHz clock with a measured 1.7 ps rms jitter at the end of a 5-bit programmable counter. Can a Butler do that? No. They use a Colpitts: (1.4meg pdf)

I have two patents that use a Butler: 3,810,234 and 4,533,881:

I gave up on the Butler in 1990 and switched to the Colpitts.

Rohde used the Butler up to about 1994, then he abandoned it and moved over to the Colpitts. He produces some of the finest low noise equipment that money can buy.

As Hajimiri and Lee have shown, the Butler is obsolete.

It's dead, Jim.

Here's an example of what can happen if you get attached to a technology and hang onto it too long: (380k wmv)

Regards,

Mike Monett

Antiviral, Antibacterial Silver Solution:

You need to look at the references. Especially the last one.

Regards,

Mike Monett

Antiviral, Antibacterial Silver Solution:

Poor Mikey. The only way you can redeem yourself is with a HAND analysis of the MC1648 that shows you actually have a single brain cell in your body. NO simulations ;-)

...Jim Thompson