Go Back   Home > Forums > >
Home Forums Rules Articles diyAudio Store Blogs Gallery Wiki Register Donations FAQ Calendar Search Today's Posts Mark Forums Read

Symetrical Colpitts oscillators
Symetrical Colpitts oscillators
Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Thread Tools Search this Thread
Old 14th November 2018, 05:32 PM   #1
Elvee is offline Elvee  Belgium
diyAudio Member
Elvee's Avatar
Join Date: Sep 2006
Default Symetrical Colpitts oscillators

Hi there,

I have already discussed the possibility of using a small oscillator as a supply source:
Simplistic Colpitts oscillators used as low-tech converters
Here is an alternative topology, possibly original.

It is based on a symetrical complementary pair:

Symetrical Colpitts oscillators-symcolp1-png

From a theoretical perspective the symmetry changes nothing, except it requires an additional active device: in AC, the transistors sex is non-relevant, meaning they are simply paralleled.
The reality is a bit more subtle though: the SE version requires an impedance, R1 in this example, to isolate the feedback node from a direct connection to the supply (this would shunt it to the GND):

Click the image to open in full size.

Ideally, this impedance would need to be an inductance, to present a zero resistance to the supply DC current and a high impedance to the feedback signal, but in order to avoid a second inductive component, a resistance is used instead.
It is a tradeoff: it needs to be small to minimize DC losses, and large to reduce the AC losses.

In real life, the resistance is made such that DC and AC losses are approximately equal.

With the symetrical version, the need for this impedance disappears, meaning an improvement in efficiency of a few points.
Of course, trading an inductor for a transistor is not particularly advantageous, but there are other benefits:
The symmetry also means a cancellation of most even-order harmonics, resulting in a better purity (and less RFI), and energy is pumped twice/cycle into the tank circuit instead of once, making the circuit better suited to higher power levels.

Initially, I had designed the circuit to feed a small LED voltmeter module that I wanted to add as an afterthought to an already built PSU.
This PSU used a small capacitive supply for the control section, and I didn't want to add a 50Hz transformer: instead, I simply inserted the new supply in series with the existing one.
It is possible, because a capacitive supply closely approximates a current source.
As the waveforms show, the circuit operates practically in class E:

Symetrical Colpitts oscillators-symcolp2-png

The "transformer" consists of 2 x 30 turns wound on a 10mm polycarbonate cylinder and separated by a thick mica washer:

Symetrical Colpitts oscillators-symcolp4-jpg

Symetrical Colpitts oscillators-symcolp3-jpg

The coupling coefficient of the face-to-face coils is pretty horrendous, ~0.45, but it works with this topology, and with this geometry, the primary to secondary common-mode capacitance of the supply (the whole circuit, not just the transformer) is a record ~3pF.
Combined with the low harmonic level, it means minimal common-mode perturbations, without any need for an additional Y capacitor.

The primary winding uses a thicker wire (0.2mm) than the secondary (0.1mm), because of the resonance current.
Without a core, the Q is bound to remain low anyway, because of the stray fields permeating the windings and generating eddy currents: any attempt to increase the Q further by increasing the wire gauge results in additional eddy currents.
The solution would be Litz wire: it would reduce the eddy currents and combat the skin effect, but it is overkill for such a simplistic circuit.

This means that without tweaking, the efficiency does not exceed ~50%, which is perfectly OK for the intended applications
Attached Images
File Type: png SymColp1.png (13.3 KB, 125 views)
File Type: png SymColp2.png (80.0 KB, 135 views)
File Type: jpg SymColp4.jpg (300.8 KB, 135 views)
File Type: jpg SymColp3.jpg (552.9 KB, 135 views)
Attached Files
File Type: asc SymColpittsReal3.asc (3.1 KB, 2 views)
. .Circlophone your life !!!! . . ♫♪ My little cheap Circlophoneİ ♫♪
Electrify your daily life!!! ....⚡⚡ ELEKTRIA ⚡⚡
  Reply With Quote
Old 14th November 2018, 07:57 PM   #2
Ketje is offline Ketje  Belgium
diyAudio Member
Ketje's Avatar
Join Date: Dec 2012
Location: Flanders
Very nice and probably original !
How about simplifing a bit ?
Attached Images
File Type: png Simpelve.png (22.5 KB, 32 views)
  Reply With Quote
Old 14th November 2018, 08:32 PM   #3
Elvee is offline Elvee  Belgium
diyAudio Member
Elvee's Avatar
Join Date: Sep 2006
Originally Posted by Ketje View Post
How about simplifing a bit ?
Regarding the resistors, it is indeed a possibility, and I considered it but rejected it in the end, at least for the BJT version.

For the MOS version I had to use it (with some adaptations) because of the different bias requirements.
-more on that later-

I used the current version for BJT's because the DC feedback from the collectors helps correcting the duty cycle and the DC centering of the "p" waveform in case of imperfect matching between the N and P transistors (imperfections which are unavoidable with real-world devices)

BTW, C2 and C4 could also be merged into a single cap, but I kept the two caps because real-world supplies have imperfections like noise, ripple or finite impedance and a bridge of capacitors is better at rejecting them.
C3 is a different story: its value is significantly lower, and the supply would need to be really dismal to require a similar splitting.
If the supply source is good, C2 and C4 can be merged without problem
. .Circlophone your life !!!! . . ♫♪ My little cheap Circlophoneİ ♫♪
Electrify your daily life!!! ....⚡⚡ ELEKTRIA ⚡⚡

Last edited by Elvee; 14th November 2018 at 08:44 PM.
  Reply With Quote
Old 15th November 2018, 06:19 PM   #4
Elvee is offline Elvee  Belgium
diyAudio Member
Elvee's Avatar
Join Date: Sep 2006
Here is the MOS version: it is based on a BSP89/BSP92 pair, but as I didn't find a quick and easy model for the BSP92, I used a BSS84 in the sim.

This means that the sim doesn't stick very well with the reality, but at least it shows how it works.

The most obvious implementation is this one, but it is rather sensitive to the threshold voltages of the MOSFETs:

Symetrical Colpitts oscillators-symcolp5-png

If the P/N thresholds are mismatched, it is necessary to adjust R4 and R5 accordingly.
For the BSP's, the matching is good.

A more robust implementation is this one: the diodes now track the saturation voltages of the MOSFETs, allowing some degree of dispersion.
This variant works very well in reality, but very poorly in sim... go figure

Symetrical Colpitts oscillators-symcolp6-png

These circuits are well suited to relatively high supply voltages, but for low-voltage operation, it is preferable to split R1 into two resistors returned directly to the supply rails.

R2 is optional, its role is to optimize the feedback phase-shift for class E operation.

Note that the real-world waveforms are much cleaner than in sim:

Symetrical Colpitts oscillators-symcolp8-jpg

Here is the test circuit, showing an alternative implementation of the "transformer":

Symetrical Colpitts oscillators-symcolp7-jpg
Attached Images
File Type: png SymColp5.png (86.6 KB, 93 views)
File Type: png SymColp6.png (74.1 KB, 90 views)
File Type: jpg SymColp7.jpg (372.4 KB, 93 views)
File Type: jpg SymColp8.jpg (306.1 KB, 107 views)
. .Circlophone your life !!!! . . ♫♪ My little cheap Circlophoneİ ♫♪
Electrify your daily life!!! ....⚡⚡ ELEKTRIA ⚡⚡
  Reply With Quote
Old 16th November 2018, 05:35 PM   #5
Elvee is offline Elvee  Belgium
diyAudio Member
Elvee's Avatar
Join Date: Sep 2006
Initially, I had designed the circuit with a specific field of application in mind: high-frequency (1~10MHz), current driven, low-power supplies based on air-core inductors, but it then occurred to me that the topology was also suited to other applications: it ticks all the right boxes as a base for clean and quiet isolated supplies.

To qualify as quiet, a supply needs to fulfill a number of criteria:
-No fast-changing waveforms, even on purely internal nodes: without drastic shielding measures, fast transitions always end up spilling somewhere.

-No injection of the operating frequency into the input or output ports; this is probably the easiest point to meet, as passive LC circuits can provide any arbitrary degree of rejection.

-Isolated grounds, because ground returns should exclusively follow their respective signals, and duplicating this connection via the PSU is bound to cause problems like ground loops.

-Good common-mode isolation between the circuits: going into the trouble of having the grounds isolated doesn't make much sense if this isolation is ruined by a large common-mode capacitance, be it inherent or added like a Y cap.
If these oscillators are made to operate at a relatively low frequency, 100kHz or less, and a low voltage 5~24V, they easily meet all of the above, thanks to their sinusoidal operation.

I made a crude test with two implementations: one based on BJT's and the other on MOSFETs.
For this occasion the transformer is now a 56µH choke with a few turns added on top to form the primary.

Symetrical Colpitts oscillators-symcolp9-png

The tuning capacitors are ordinary X7R MLCCs and the transistors are BD438/439.

Symetrical Colpitts oscillators-symcolp10-jpg

The result is surprisingly good, considering the simplicity of the setup, and its construction on a breadboard: the contact resistances have a huge effect, and if you touch any component you see the amplitude vary.
The circuit even tolerates E-caps for the larger value capacitors.

Symetrical Colpitts oscillators-symcolp11-jpg

Here is the MOS variant:

Symetrical Colpitts oscillators-symcolp12-png

The physical test circuit is somewhat different, because the best PMOS of my stock is a MTP23P06, not a very good match for the IRFZ44 I used on the other side. As a result, I had to alter the bias resistors values, but even so, a significant asymmetry remains.

Symetrical Colpitts oscillators-symcolp13-jpg

Symetrical Colpitts oscillators-symcolp14-jpg

It still manages to work though.
Of course, in a real implementation, the transformer doesn't need to be that crude: it could be a properly gapped ferrite core or a metal powder toroid.
Attached Images
File Type: png SymColp12.png (85.0 KB, 64 views)
File Type: jpg SymColp13.jpg (450.1 KB, 70 views)
File Type: jpg SymColp14.jpg (285.6 KB, 69 views)
File Type: png SymColp9.png (67.6 KB, 67 views)
File Type: jpg SymColp10.jpg (480.9 KB, 64 views)
File Type: jpg SymColp11.jpg (212.0 KB, 89 views)
. .Circlophone your life !!!! . . ♫♪ My little cheap Circlophoneİ ♫♪
Electrify your daily life!!! ....⚡⚡ ELEKTRIA ⚡⚡
  Reply With Quote


Symetrical Colpitts oscillatorsHide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump

Similar Threads
Thread Thread Starter Forum Replies Last Post
Simplistic Colpitts oscillators used as low-tech converters Elvee Power Supplies 17 26th December 2016 09:40 AM
FS: Symetrical B1 Buffer by Salas dantwomey Swap Meet 2 17th May 2010 12:47 PM
Symetrical out low Zout karsten21 Tubes / Valves 5 1st February 2010 11:19 PM
going balanced/symetrical - what benefits? weissi Solid State 15 20th October 2007 09:06 AM
Symetrical field, is this? Raka Multi-Way 6 14th September 2003 02:21 PM

New To Site? Need Help?

All times are GMT. The time now is 07:49 PM.

Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2019 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 15.00%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2019 DragonByte Technologies Ltd.
Copyright ©1999-2019 diyAudio