Sizing a variable frequency drive is not hard. Getting from the size to something a distributor can quote is where it falls apart, because the two halves of the job live in different documents. The sizing tables are in the design guide. The ordering grammar is in the catalog. Most sizing calculators stop at an amp figure, and most catalogs assume you already know the amps.
This guide covers the whole path: what to read off the motor nameplate, why amps beat horsepower once you have both, how the four major brands label duty differently and what that costs you when you compare them, what changes when the supply is single phase, and why the string a sizing tool hands you is sometimes a complete catalog number and sometimes only a fragment. It is written against the same rating dataset the VFD Part Number Finder runs on, so the counts and ranges quoted here are the ones the tool will actually use.
1. Size on Full-Load Amps, Not Horsepower
Horsepower is how drives are marketed and how most people think about motors. Current is what the drive has to survive. Those are not the same thing, and the gap between them is wide enough to change the answer.
Two facts make horsepower an unreliable key. First, the kilowatt and horsepower ladders are different ladders: a 7.5 HP motor and a 5.5 kW motor are the same machine, but the arithmetic conversion gives 5.59 kW, which reads as "the drive is smaller than the motor" against a published 5.5 kW rating and pushes every correctly sized drive out of the answer. The finder handles that by snapping an entered horsepower down onto the nearest published motor step when the gap is inside 8 percent, which reproduces the standard IEC and NEMA pairs (7.5 HP with 5.5 kW, 15 HP with 11 kW, 25 HP with 18.5 kW, 50 HP with 37 kW) without inventing a value. Second, brands publish different output currents for what they each call the same motor size. A 10 to 15 percent spread between equivalent drives is normal and comes purely from rating convention.
So the working order is: read the full-load amps off the nameplate, pick the duty, and require the drive rated output current at that duty to meet the amps. Horsepower then becomes a sanity check on the motor label rather than the selection key. The finder does exactly this, and when it has both figures it raises a flag if the drive rating and the entered motor size disagree by more than a quarter, because that usually means one of the two numbers was read off the wrong nameplate.
2. ND, HD, VT, CT, LD: The Same Idea Under Five Names
Every drive carries more than one rating because a fan and a conveyor ask different things of it. A variable-torque load draws its full current only at full speed, so the drive can be rated closer to its thermal limit with a modest short-term overload allowance. A constant-torque load can demand full current at low speed, where the drive fan and heatsink work harder and one output device carries the current for longer, so the same hardware is derated and given a larger overload allowance.
The naming is where cross-shopping goes wrong. Normal Duty and Heavy Duty (ND and HD) is the most common pair. Variable Torque and Constant Torque (VT and CT) describes the same split from the load side rather than the drive side. Some catalogs add Light Duty (LD) above Normal Duty for the least demanding fan and pump work, giving a three-step ladder. Danfoss design guides use Normal Overload and High Overload for the two columns. None of these are interchangeable across a table boundary: the useful question is always "how many amps continuously, and how much overload for how long," and that is what to compare.
What each brand actually publishes in the finder dataset matters as much as what it calls it:
| Brand | What the rating data carries |
|---|---|
| ABB ACS | Separate normal-duty and heavy-duty current on 275 of 305 rows. The ACS355 rows are the exception and carry the normal-duty column only. |
| Yaskawa | Separate normal-duty and heavy-duty current on every one of 187 rows. |
| Allen-Bradley PowerFlex | Normal-duty current only, on all 419 rows. |
| Danfoss VLT | One continuous rating. The second column in the source is an intermittent overload peak, not a continuous heavy-duty rating, so it is deliberately not used. |
The practical consequence: ask for heavy duty and the ABB and Yaskawa answers are read from a real heavy-duty column, while the PowerFlex and Danfoss answers are read from the continuous column and flagged. The catalog heavy-duty size for those two is usually a frame or two smaller than the continuous figure shown, so a constant-torque job needs that number confirmed in the catalog before the order goes out.
3. Single Phase, Three Phase, and the 400 V Class Trap
Two separate things get called "single-phase drive," and confusing them is expensive. The first is a drive the manufacturer publishes for a single-phase supply, with its own rating table. The second is a three-phase-input drive fed from single-phase power, which is a derating problem, not a catalog lookup. The finder covers the first and explicitly refuses the second.
Drives published for a single-phase supply exist in all four brands in this dataset, but they are a small and low-power corner of it. There are 25 PowerFlex rows (PowerFlex 523 and 525, at 100 to 127 V and 200 to 240 V), 33 ABB rows (ACS180, ACS355, ACS380, ACS480, at 208 to 240 V), 18 Danfoss rows (FC 51 and FC 202 at 200 to 240 V, plus the four FC 202 rows at 380 to 480 V), and 14 Yaskawa rows (GA500 and V1000, at 200 to 240 V). Most of those families stop between 2.2 and 3.7 kW on a single-phase supply, which is exactly why the "double the horsepower" folk rule exists for feeding a three-phase drive from single-phase power. The one outlier in this dataset is the Danfoss AQUA Drive FC 202, whose nine single-phase 200 to 240 V rows run from 1.1 kW all the way to 22 kW at 88 A, and which also fields the only single-phase 380 to 480 V line here: four rows from 7.5 to 37 kW, straight from the AQUA Design Guide mains-supply table. That folk rule is a different calculation from anything this dataset holds and belongs on its own page; do not read a three-phase rating row as usable on single-phase power because the output current looks big enough.
The voltage trap is subtler and catches more people. Inside the 380 to 480 V band, the class label is not the rating. Some rows are published at 380 to 480 V and genuinely cover a 480 V plant. Others are published at 380 to 415 V, the European 400 V unit, and do not, even though both get called 400 V class. In this dataset the trap is live on ABB, which carries 380 to 415 V rows alongside its 380 to 480 and 380 to 500 V ones, and on PowerFlex, which carries 380 to 415, 380 to 480, and 440 to 480 V rows; every Yaskawa row in the band is published at 380 to 480 V, per the GA500, GA800, A1000, and V1000 technical manuals, and every Danfoss row at 380 to 480 or 380 to 500 V. The finder ranks a drive that truly covers your supply above one that only comes close, and flags the near-miss rather than hiding it. Read the published range, not the class label.
4. Why a Rating Code Is Not a Part Number
This is the part that separates a sizing answer from something you can put on a purchase order, and it is the reason a lot of drive quotes come back wrong.
A drive ordering string is built from segments. Some of them follow from the motor: the rating and the voltage class. The rest do not, and no amount of motor data will produce them. Mounting style, EMC filter fitted or not, enclosure type, keypad, fieldbus card, I/O, and safety options are ordering choices made by whoever specifies the panel. A sizing table publishes the rating segment because that is what the table is about. The catalog publishes the whole grammar.
In the finder dataset the split falls this way. Allen-Bradley PowerFlex rows carry a complete orderable catalog number, and are labelled as such only when the string matches one of the two published Rockwell catalog shapes with no open segment left. ABB rows carry the rating segment: an ABB type designation runs series, then construction, then rating, then voltage class, and the construction segment covering mounting style, EMC filter, and variant is not derivable from a motor rating. Yaskawa rows carry the rating key, which the full model number wraps with the series prefix and the enclosure, EMC, and option digits. Danfoss rows carry the power code, which the full VLT ordering string wraps with the series prefix, voltage class, enclosure, EMC filter, and option selections.
Every row in the tool says which of the two it is and what a full ordering string still needs, and it never assembles or edits a string to look more complete than the source allows. The correct next step for a rating-code row is to take the rating to the manufacturer selection guide or your distributor and finish the string there. That is one phone call. A guessed segment on a purchase order is a returned drive and a lost week.
5. Worked Example: 7.5 HP, 480 V, Three Phase, Normal Duty
A 7.5 HP fan motor on a 480 V three-phase supply, variable torque, so normal duty. Entering 7.5 HP converts arithmetically to 5.59 kW, which snaps down to the published 5.5 kW motor step, and the search runs in the 380 to 480 V band.
The PowerFlex answer is the one worth showing in full, because it is the one that arrives as a complete catalog number: 25B-D013N104, a PowerFlex 525 rated 13 A at normal duty, 5.5 kW and 7.5 HP, frame C, published range 380 to 480 V. That string is verified against the Rockwell product page and pinned by a unit test in the app, so it does not drift with a data rebuild.
The other three brands return rating rows in the same envelope rather than orderable strings, which is the honest result rather than a worse one. The useful reading of that table is not "which brand won" but "what is the closest rating in each brand, and what does each one still need from me." For PowerFlex the answer is the enclosure type and any option accessories. For ABB, Yaskawa, and Danfoss it is the construction or option segments listed in the section above. If the load had been a conveyor rather than a fan, the same query at heavy duty would read the ABB and Yaskawa heavy-duty columns directly and flag the PowerFlex and Danfoss rows as continuous-rating matches that need a catalog check.
6. PowerFlex 525 vs GA500
These two land in the same slot in most plants: a compact general-purpose drive for pumps, fans, conveyors, and small machinery. In the rating data they cover a similar envelope with a different shape. In the 380 to 480 V three-phase band, the PowerFlex 525 dataset holds 20 rating rows spanning 0.4 to 18.5 kW normal duty and 1.4 to 43 A, all published at 380 to 480 V. The GA500 dataset holds 13 rows in the same band spanning 0.4 to 30 kW and 1.2 to 60 A, also published at 380 to 480 V per the GA500 technical reference. Both families genuinely cover a 480 V North American plant, so voltage does not separate them; where the envelopes differ is that the GA500 rows reach higher power in this band.
The second material difference is what each publishes about duty and about ordering. Every GA500 row in this dataset carries a separate heavy-duty current alongside the normal-duty one, so a constant-torque selection is read from a real heavy-duty column. PowerFlex rows carry the normal-duty column only, so the same constant-torque request is matched on the continuous figure and flagged for a catalog check. Running the other way, PowerFlex 525 rows arrive as complete Rockwell catalog numbers you can quote directly, while GA500 rows arrive as rating keys that still need the series prefix and the enclosure, EMC, and option digits. So the split, in plain terms: if the load is constant torque and the sizing margin is tight, the GA500 data answers the question you are actually asking; if you want an orderable string in one step, the PowerFlex 525 data does. As general engineering practice rather than anything in this data, plants standardized on one automation vendor usually have a strong reason to stay there for spares, drive-to-controller integration, and technician familiarity, and that is a legitimate tiebreaker when the electrical fit is equivalent.
7. PowerFlex 525 vs ACS355
This pairing shows up in older specifications and in quotes from surplus channels, so the first thing to say about it is a lifecycle fact rather than a spec: the ACS355 is flagged in the family data as a legacy line superseded by the ACS380. It is still shown in results because a drive you have to match today does not stop existing when it is superseded, but it is deliberately ranked below current products when one fits equally well. If you are specifying a new panel, the like-for-like comparison to run is PowerFlex 525 against ACS380 or ACS480, not against ACS355.
On the numbers, the two are close in the 380 to 480 V three-phase band. PowerFlex 525 holds 20 rows spanning 0.4 to 18.5 kW and 1.4 to 43 A; ACS355 holds 14 rows spanning 0.37 to 22.0 kW and 1.2 to 44.0 A. Both are published across the full 380 to 480 V range, so both genuinely cover a 480 V supply and the voltage question never arises. Two differences remain. First, duty data: the ACS355 rows are the one ABB family in this dataset with no separate heavy-duty current, so unlike ACS380, ACS480, ACS580, and ACS880 they behave like the PowerFlex rows and give a continuous-rating answer that needs a catalog check for constant-torque work. Second, ordering: PowerFlex 525 rows are complete catalog numbers, while ACS355 rows are rating segments that still need the construction segment covering mounting style, EMC filter, and variant. Picking between them on the electrical envelope alone is close to a coin flip in the overlapping range; picking on lifecycle is not, and that is the axis that should decide it for a new installation.
8. GA500 vs Danfoss VLT
Danfoss is not one line, which is the first thing this comparison has to sort out. In the 380 to 480 V three-phase band the dataset holds five VLT families: the Micro Drive FC 51 (12 rows, 0.37 to 22 kW, 1.2 to 43 A), the HVAC Drive FC 102 (17 rows, 1.1 to 90 kW), the AQUA Drive FC 202 (10 rows, 0.37 to 7.5 kW), and the AutomationDrive FC 301 and FC 302 (38 rows each, 0.37 to 75 kW, 1.2 to 147 A, published at 380 to 480 V and 380 to 500 V). Against a GA500 at 13 rows and 0.4 to 30 kW, the honest statement is that the AutomationDrive families cover a materially wider power range, while the GA500 rows in this dataset are published at 380 to 480 V and cover a 480 V supply just as the AutomationDrive rows do. The FC 51 sits at the compact end and overlaps the GA500 most directly on power.
The trade runs the other way on duty. Every GA500 row carries a published heavy-duty current; every Danfoss row here carries one continuous rating, because the second column in the Danfoss source is an intermittent overload peak rather than a continuous heavy-duty rating and using it would overstate what the drive can do all day. So a constant-torque selection against Danfoss is a continuous-rating match plus a catalog check, and against GA500 it is a direct read. One coverage caveat belongs on the record: Danfoss FC 202 coverage in this dataset is deliberately incomplete because 39 rows with a suspect high-overload key were dropped at bake time rather than shipped wrong, and FC 301 and FC 302 cover the same envelope with correct data. Both sides return rating segments rather than orderable strings, so neither wins the ordering step. As general engineering practice rather than a claim from this data, the FC 302 is positioned as the dynamic and high-performance member of the AutomationDrive pair while FC 301 is the general-purpose one, which is the split to keep in mind when the load is a hoist or a high-dynamic machine axis rather than a pump.
9. Working the Result Without Getting Burned
A sequence that keeps the judgment where it belongs:
- Enter amps if you have them. Nameplate full-load amps switch the tool onto the hard-limit basis. Horsepower or kilowatts alone still work, and the result says the current check was not run.
- Set the duty. Variable torque for fans and centrifugal pumps, constant torque for conveyors, crushers, hoists, and positive-displacement pumps. An unset duty is a normal-duty answer.
- Clear the voltage flags. Any row whose published range does not include your supply is a pointer, not a fit. Confirm the range in the current manufacturer catalog for your market.
- Read the completeness label. A complete catalog number can be quoted. A rating code needs the construction, enclosure, EMC, and option segments finished with the manufacturer or distributor.
- Confirm the heavy-duty size where the data is one column. For PowerFlex and Danfoss, a constant-torque job needs the catalog heavy-duty rating, which is usually a frame or two smaller than the continuous figure.
- Close the gap the tool cannot see. Physical fit and mounting, terminal layout and cable entry, branch-circuit protection and line reactors, harmonic limits, controls and fieldbus, functional safety level and approvals, ambient and altitude derating, and whether the existing feeder and starter wiring suit the drive you picked.
- Send the RFQ block, not a summary. The copy-out block carries the supply, motor data, duty, candidate strings, and the completeness labels together, which is what an inside sales desk needs to quote without a second email.
Done in that order the tool saves you the catalog cross-walk and leaves the decisions with the person who knows the installation, which is the only place they can safely live.