Tweet
Based on JL Kings valuable observations and simulation schematics. Found here post #1479.
https://www.geotech1.com/forums/show...etector/page60
I was able to develop some working circuits. The gate drive transformers and power supply are the biggest challenge. Like the Johnson regenerative half bridge Tx, the vmh3cs generates its own virtual ground at 1/2 vcc. The virtual (Tx) ground is delicately balanced and will not handle additional loads. So the system ground needs to be buffered and referenced to Tx ground or 1/2 vcc. The gate drive transformers are a study in themselves. These provide isolation, level shifting along with a boosted gate drive voltage. This allows the use of high side n-ch mosfets. The best cores I tested were made by Magnetics Inc. "Square Permalloy" material code D. These exhibit a square B-H curve. Fourteen turns trifilar magnet wire, one wire as primary and the other two connected as a center tapped secondary with a 1:2 turns ratio. These cores are rare, large and expensive. I had a variety from past projects to test. Next I tried surface mount gate drive transformers from Coilcraft, these failed as they are low inductance, low volts/us product. They are meant to be used in pwm drives with middle range duty cycles or line drivers for data communications. For experimenting I found common mode choke cores to be economic and available. I ordered a variety of "common mode choke toroids" from ebay. They come bifilar wound and are used in switch mode power supplies as filters. You can get 10 for under $2. The same core can be bought without windings. The permeability is on the high side, but they work and are easy add or remove turns. No specs are available for the original vmh3cs core or the ebay core. One comparison is the vmh3 core measured about 12uh/turn while the green cores have about 100uH/t. When the next pcbs come in I'm going to tweak the turns for best performance. Also need to test these core for saturation. In general cores should be a "soft ferrite" with a permeability around 2000.
The attached scope shots are - green trace equals coil voltage, yellow equals current across a 0.5 ohm resistor. At turn on, over 200 volts is discharged into the tx coil. In a few uS the current is almost a constant 2 amps. The active damping is the key to the vmh3. Tests without active damping showed a fast coil (340uh 0.6r) generated 260vpp 1 amp and a slow (1500uh 3r) coil produced about 115vpp and less than one amp. Active damping applied at the right time during flyback caused a huge increase in current and voltage for fast and slow coils. For fast coils the damping must be applied earlier. The slow coil damped at 5us created a coil voltage of 800vpp, in other words the high and low side mosfets were in avalanche. I added a 240v snubber to limit the vpp to 480v. Also reducing the supply voltage can control the output. Reading the Vallon patent, they monitor the coil current to control the peak output. The timing for the active damping explains why the vmh3cs only likes a small range of coils. A fast coil is already in decay before the damping resistor is connected at 4.5us. Opposite if the damping is activated too soon on a slow coil the fly back gets snubbed before it reaches full potential.
The circuits were tested with a supply voltage from 12 to 20 volts, supply current spanned 100mA to 350mA depending on the coil specs and applied voltage. Switching logic requires 8 I/O lines , all switching pulses are 5v, 2us wide. The Tx pulse width was variable 30uS to 100uS.
https://www.geotech1.com/forums/show...etector/page60
I was able to develop some working circuits. The gate drive transformers and power supply are the biggest challenge. Like the Johnson regenerative half bridge Tx, the vmh3cs generates its own virtual ground at 1/2 vcc. The virtual (Tx) ground is delicately balanced and will not handle additional loads. So the system ground needs to be buffered and referenced to Tx ground or 1/2 vcc. The gate drive transformers are a study in themselves. These provide isolation, level shifting along with a boosted gate drive voltage. This allows the use of high side n-ch mosfets. The best cores I tested were made by Magnetics Inc. "Square Permalloy" material code D. These exhibit a square B-H curve. Fourteen turns trifilar magnet wire, one wire as primary and the other two connected as a center tapped secondary with a 1:2 turns ratio. These cores are rare, large and expensive. I had a variety from past projects to test. Next I tried surface mount gate drive transformers from Coilcraft, these failed as they are low inductance, low volts/us product. They are meant to be used in pwm drives with middle range duty cycles or line drivers for data communications. For experimenting I found common mode choke cores to be economic and available. I ordered a variety of "common mode choke toroids" from ebay. They come bifilar wound and are used in switch mode power supplies as filters. You can get 10 for under $2. The same core can be bought without windings. The permeability is on the high side, but they work and are easy add or remove turns. No specs are available for the original vmh3cs core or the ebay core. One comparison is the vmh3 core measured about 12uh/turn while the green cores have about 100uH/t. When the next pcbs come in I'm going to tweak the turns for best performance. Also need to test these core for saturation. In general cores should be a "soft ferrite" with a permeability around 2000.
The attached scope shots are - green trace equals coil voltage, yellow equals current across a 0.5 ohm resistor. At turn on, over 200 volts is discharged into the tx coil. In a few uS the current is almost a constant 2 amps. The active damping is the key to the vmh3. Tests without active damping showed a fast coil (340uh 0.6r) generated 260vpp 1 amp and a slow (1500uh 3r) coil produced about 115vpp and less than one amp. Active damping applied at the right time during flyback caused a huge increase in current and voltage for fast and slow coils. For fast coils the damping must be applied earlier. The slow coil damped at 5us created a coil voltage of 800vpp, in other words the high and low side mosfets were in avalanche. I added a 240v snubber to limit the vpp to 480v. Also reducing the supply voltage can control the output. Reading the Vallon patent, they monitor the coil current to control the peak output. The timing for the active damping explains why the vmh3cs only likes a small range of coils. A fast coil is already in decay before the damping resistor is connected at 4.5us. Opposite if the damping is activated too soon on a slow coil the fly back gets snubbed before it reaches full potential.
The circuits were tested with a supply voltage from 12 to 20 volts, supply current spanned 100mA to 350mA depending on the coil specs and applied voltage. Switching logic requires 8 I/O lines , all switching pulses are 5v, 2us wide. The Tx pulse width was variable 30uS to 100uS.
Comment